Quill OS - Kobo Nia model A (N306A)

@Szybet , here's a porting thread to get your Nia set up properly. I usually don't have much time for InkBox except on week-ends, but you can checkout the Libra H2O porting thread to have an idea of what will need to be done.

4:34 AM

I do not know if @andi has done anything with mx6ull yet (Nia's SoC), but it sure could be interesting to have mainline there too, if it's not too much work.

I have already the backup of the SD card

8:26 AM

i dont understand the openssl thing that was there talked about (edited)

8:26 AM

i will get to u-boot today

8:27 AM

and it would be great to make a 5.x kernel work

https://github.com/Kobo-InkBox/kernel/blob/master/initrd/common/overlay-mount Maybe look here, it's how the rootfs is verified and mounted at boot (it's a squashfs archive)

kernel/overlay-mount at master · Kobo-InkBox/kernel

Kernel sources for InkBox OS. Contribute to Kobo-InkBox/kernel development by creating an account on GitHub.

hmm i understand the script

3:05 PM

but i dont know how it could help

How what could help?

tux-linux

https://github.com/Kobo-InkBox/kernel/blob/master/initrd/common/overlay-mount Maybe look here, it's how the rootfs is verified and mounted at boot (it's a squashfs archive)

this

lets focus on the kernel / bootloader first?

4:20 PM

or there doesnt need to be anything changed in them?

Um yeah

5:02 PM

U-Boot kernel sector Change it to 81920 And change kernel load size to 18432

hmm why is it needed to change?

5:03 PM

so i need to compile u-boot?

5:06 PM

and, is it all?

Szybet

hmm why is it needed to change?

Kernel is too big

why is it too big?

Do something like this https://github.com/Kobo-InkBox/kernel/commit/a6754443ad14f9033d7e328426c9ad4451f516b7

N437: Change kernel load sector & size · Kobo-InkBox/kernel@a675444

whot would change that it would be big?

It can't fit in the default unpartitioned space

5:19 PM

Initrd is embedded

it would be better to understand if you would provide the full sd card partition / place plan

Download factory image and look

5:21 PM

23.163.0.39

where is it

and, maybe we try to update the kernel to a newer version?

See here for example http://23.163.0.39/bundles/inkbox/native/n905b/inkbox-1.7-n905b.xz

i will look at it

how do i generate the u-boot config?

7:04 PM

or even get the board name

Szybet

it would be better to understand if you would provide the full sd card partition / place plan

inkbox-1.7-n905b: DOS/MBR boot sector; partition 1 : ID=0x83, start-CHS (0x2,140,10), end-CHS (0x4,247,51), startsector 40959, 38913 sectors; partition 2 : ID=0x83, start-CHS (0x6,127,58), end-CHS (0x46,61,55), startsector 104448, 1024000 sectors; partition 3 : ID=0x83, start-CHS (0x46,61,56), end-CHS (0x56,142,56), startsector 1128448, 262144 sectors; partition 4 : ID=0x83, start-CHS (0x56,142,57), end-CHS (0x1d5,242,17), startsector 1390592, 6159156 sectors

7:28 PM

i used file to show this

7:28 PM

do i need to mount this, or use another app to open this?

7:29 PM

the easiest and fastest way would be if you shared resources / links how to do those steps

7:30 PM

maybe write a guide / documentation?

fdisk img

7:38 PM

Or xzcat img > InkBox.img

7:39 PM

losetup -P -f --show InkBox.img

7:39 PM

fdisk -l /dev/loop<x>

Szybet

how do i generate the u-boot config?

I don't know the kernel sources or U-Boot sources for this model, but look in include/configs for U-Boot

7:46 PM

Maybe imx_v7_defconfig

tux-linux

losetup -P -f --show InkBox.img

thats whot i thought, that i need to mount it

tux-linux

I don't know the kernel sources or U-Boot sources for this model, but look in include/configs for U-Boot

ok i will have a look at it

1

Maybe you can set up the emulator in a Debian VM to have an idea of what you will be dealing with (edited)

12:20 AM

https://github.com/Kobo-InkBox/emu

GitHub - Kobo-InkBox/emu: InkBox OS emulator

InkBox OS emulator. Contribute to Kobo-InkBox/emu development by creating an account on GitHub.

1:19 PM

message.txt

2.69 KB

1:19 PM

this looks promising

What filename?

auto.conf

Doubt it will work

tux-linux

Maybe imx_v7_defconfig

it isin't there

5:55 PM

but i have a idea

5:55 PM

the cpu is mx6ull

yes

% ls | grep mx6ull
mx6ull_14x14_ddr3_arm2_defconfig
mx6ull_14x14_ddr3_arm2_emmc_defconfig
mx6ull_14x14_ddr3_arm2_epdc_defconfig
mx6ull_14x14_ddr3_arm2_nand_defconfig
mx6ull_14x14_ddr3_arm2_qspi1_defconfig
mx6ull_14x14_ddr3_arm2_spinor_defconfig
mx6ull_14x14_ddr3_arm2_tsc_defconfig
mx6ull_14x14_evk_defconfig
mx6ull_14x14_evk_emmc_defconfig
mx6ull_14x14_evk_nand_defconfig
mx6ull_14x14_evk_qspi1_defconfig
mx6ull_9x9_evk_defconfig
mx6ull_9x9_evk_qspi1_defconfig
mx6ull_ntx_lpddr2_256m_defconfig
mx6ull_ntx_lpddr2_256m_mfg_defconfig
mx6ull_ntx_lpddr2_512m_defconfig
mx6ull_ntx_lpddr3_512m_defconfig
mx6ull_ntx_lpddr3_512m_mfg_defconfig

How much RAM do you have in the Nia?

wait, i will check

5:56 PM

RAMType='LPDDR2

5:57 PM

RamSize='256MB'

5:57 PM

so its only

mx6ull_ntx_lpddr2_256m_defconfig
mx6ull_ntx_lpddr2_256m_mfg_defconfig

5:58 PM

whot does mfg mean

Kobo still has done the mistake to put 256MB of RAM in a 2 years old device

5:58 PM

Don't use mfg

whot does mfg mean

mfg=manufacturing I guess

5:58 PM

I never use them

tux-linux

Kobo still has done the mistake to put 256MB of RAM in a 2 years old device

you know, nia is cheap

Just take mx6ull_ntx_lpddr2_256m_defconfig

and why would it need more?

I do know

its a ereader

Well, Elipsa (2021) has 1G, Glo HD (2015) has 512M, Libra (2019) has 512M, Clara (2018) has 512M, so

5:59 PM

Also there is a substantial speed increase with InkBox devices which have 512M of RAM (edited)

5:59 PM

But anyways, end of the rant

nia is just cheap

6:02 PM

it just writted it to the config?

6:02 PM

now i need to select whot i need? ohn o...

Yes

6:02 PM

No

whot no

Remove the mx6ull bit

tux-linux

Remove the mx6ull bit

?

And it will compile u-boot.bin

6:03 PM

make -j6 CROSS_COMPILE=arm-linux-gnueabihf- ARCH=arm

message.txt

22.69 KB

6:03 PM

hahahaha...

Try with a different compiler

6:04 PM

I cannot help you on those errors

6:04 PM

But you can build this code somehow

6:04 PM

Sometimes just changing compiler will solve the issue

tux-linux

Try with a different compiler

like which one? i have the 4.9

tux-linux

Sometimes just changing compiler will solve the issue

i dont think changing to a older one will help

Did you clone the InkBox kernel repo?

Szybet

i dont think changing to a older one will help

it did for me

6:05 PM

changing to a newer one

im using the compiler from kobo reader repo

Ok; clone InkBox kernel repo

6:05 PM

https://github.com/Kobo-InkBox/kernel

GitHub - Kobo-InkBox/kernel: Kernel sources for InkBox OS

Kernel sources for InkBox OS. Contribute to Kobo-InkBox/kernel development by creating an account on GitHub.

6:05 PM

And use arm-nickel-linux-gnueabihf- If it doesn't work, try armv7l-linux-musleabihf-

6:06 PM

Or maybe you will have more luck with bootstrapping this one https://github.com/koreader/koxtoolchain

GitHub - koreader/koxtoolchain: Build script for various toolchains...

Build script for various toolchains used by KOReader - GitHub - koreader/koxtoolchain: Build script for various toolchains used by KOReader

i will

1

If you really aren't lucky, I'll try to compile it myself

6:08 PM

Sometimes there are Unexplainable Build Errors™️ that are solved just by compiling on another machine or distro

C and other c are something shitty

6:27 PM

but i got it

Nice!!

6:28 PM

What toolchain?

tux-linux

What toolchain?

mine, 4.9

6:28 PM

yours doesn't worked

6:28 PM

but the problem was a package i have installed

Ah

https://github.com/openbmc/openbmc/issues/3274

Build error using Debian testing in libfdt · Issue #3274 · openbmc/...

Hi, I'm attempting to build OpenBMC on Debian testing. This is the command that fails: ocp-bluecmd:~/openbmc/build/tmp/work/x86_64-linux/u-boot-mkimage-native/1_2017.09-r0/git$ gcc -isystem...

6:28 PM

can i write it using dd in kobo or do i need to get the sd card from inside?

6:29 PM

and write it using the PC

you can use the Kobo

6:29 PM

dd if=u-boot.bin of=/dev/mmcblk0 bs=1K seek=1 skip=1

tux-linux

dd if=u-boot.bin of=/dev/mmcblk0 bs=1K seek=1 skip=1

actually im not sure if this is the correct command

6:49 PM

here is a part of the script from the device, the update script

if [ `echo $UBOOT | grep -c imx` == 1 ]; then
        dd if=$UBOOT of=/dev/$DEVICE bs=1K seek=1
    else
        dd if=$UBOOT of=/dev/$DEVICE bs=1K seek=1 skip=1

6:49 PM

i think imx is a procesor type

Maybe omit the skip then

6:49 PM

Look in the update zips from Kobo

6:49 PM

See what it's called

mine nia has mx6ull

tux-linux

See what it's called

it called using cpu names

6:50 PM

so i will not use skip

6:50 PM

no, i think wrong

6:50 PM

i will, there is a else

backup the boot sector

6:51 PM

what's fdisk -l on your device?

tux-linux

backup the boot sector

i have backup of the whole sd card

6:52 PM

like, i bricked my ereader 5 times already

ok, well flashing only the boot sector would probably be better for the sd

i can copy only the boot sector from the sd card

Ok

6:53 PM

Well, do that if you brick it

its dead

6:55 PM

D:

6:55 PM

even the led doesn't work

then probably skip=1

XD

6:58 PM

my pc doesn't list partitions?

6:59 PM

um...

You probably just have overwritten the DOS boot sector!

6:59 PM

Congrats

6:59 PM

So fdisk is like what in the world has happened to this thing

??? how

Just reflash the unpartitioned space, you'll get it back

7:00 PM

Did you include skip or seek?

i included seek

strange then

7:00 PM

idk

partitions are back

7:06 PM

u-boot works

7:07 PM

but the kernel does not boot

Oh...

oh i know

Strange

i had a custom one

Ah

i randomly copied 5 Mb of the boot sector

7:07 PM

i think i cut the kernel in half

7:07 PM

and the other half is the custom one

Whoops

30 Mb will be good then :D

the second option does not work either, but now i see the partition table

7:20 PM

so u-boot does not work

7:20 PM

hmm, maybe the config?

I the cmd in the oryginal u-boot, there were those commands, in the config they are off, so the config is wrong

7:39 PM

config might be wrong, yes Kobo doesn't always release the atomic copy of their sources trees

tux-linux

config might be wrong, yes Kobo doesn't always release the atomic copy of their sources trees

this is not cool

7:40 PM

its very bad

yes it is

can i launch u-boot from u-boot?

yes

so i dont need to rewrite it every time

I think you need to load it to memory first, then go <memory address>

7:41 PM

But maybe you can use download mode

i will download it using tftp

Is there a download_mode command in your U-Boot?

i checked the command

7:42 PM

there were many of them, like tftp, usb, i2c, boot from usb etc

You have to use uuu to download U-Boot

7:42 PM

https://github.com/NXPmicro/mfgtools

GitHub - NXPmicro/mfgtools: Freescale/NXP I.MX Chip image deploy to...

Freescale/NXP I.MX Chip image deploy tools. Contribute to NXPmicro/mfgtools development by creating an account on GitHub.

whots that?

An USB downloader

7:43 PM

Takes U-Boot and writes it into the Nia's RAM

7:43 PM

Then starts it

which command does this to ram?

uuu u-boot.imx

7:44 PM

with sudo if required

and i think i can fastboot boot img?

Yes, but that is for the kernel only AFAIK

tux-linux

uuu u-boot.imx

this command has not a good description

If there is no download_mode command in U-Boot, then you could enter this:

eBR-1A # mw 20d8040 30
eBR-1A # mw 20d8044 10000000
eBR-1A # reset

But I suggest you wait a bit; @andi might have some warnings since it's a different platform

Szybet

this command has not a good description

Basically does "Take u-boot.imx file and load it in IMX USB device plugged in, then start it" (edited)

tux-linux

If there is no download_mode command in U-Boot, then you could enter this:

eBR-1A # mw 20d8040 30
eBR-1A # mw 20d8044 10000000
eBR-1A # reset

But I suggest you wait a bit; @andi might have some warnings since it's a different platform

?... whots that

Basically download_mode without the actual command

:|

Writes to memory addresses

tux-linux

Basically does "Take u-boot.imx file and load it in IMX USB device plugged in, then start it" (edited)

if it fails, it will reset?

7:48 PM

i mean the u-boot

7:48 PM

or i can just reboot it

7:48 PM

nevermind

So that the SoC thinks there's a download mode pin activated until a complete hardware reset is done (power down, power up, not just reset in U-Boot)

Szybet

if it fails, it will reset?

No, you have to do the power-down-up sequence described before

7:49 PM

But I suggest we wait for andi1 to give some advice Maybe it could physically damage your device, I don't know (edited)

7:49 PM

Otherwise, if you have the download_mode command, you can try that

tux-linux

If there is no download_mode command in U-Boot, then you could enter this:

eBR-1A # mw 20d8040 30
eBR-1A # mw 20d8044 10000000
eBR-1A # reset

But I suggest you wait a bit; @andi might have some warnings since it's a different platform

i will not use those command, they feel dangerous

Well, we have to know if we're writing the right thing at the right place, after that we don't have to worry

7:50 PM

Maybe you can try the tftp thing you were talking about

i dont know if a u-boot kernel will boot when it expects a normal kernel

7:51 PM

but u-boot is linux too

7:51 PM

¯\_(ツ)_/¯

7:53 PM

does u-boot have wifi support?

7:53 PM

nope

Szybet

but u-boot is linux too

no

7:56 PM

maybe you can load it on serial

7:56 PM

loady

7:56 PM

loadx

7:56 PM

loadb

7:56 PM

whatever

tux-linux

no

i readed smth in a wrong way

7:59 PM

i tryied booting it using fastboot

7:59 PM

data abort

7:59 PM

whot does it mean

8:13 PM

and this hangs

I honestly don't know

Szybet

Click to see attachment 🖼️

Is the Kobo put in download mode?

8:26 PM

(not fastboot)

you mean a command named download_mode

8:27 PM

?

8:33 PM

8:34 PM

now, how do i launch it

Szybet

you mean a command named download_mode

yes

this command is missing

8:35 PM

i have those options

Szybet

now, how do i launch it

Honestly, I do not know for sure Maybe go 0x80800000, assuming it's the address where it put the U-Boot

8:36 PM

its there

8:36 PM

i will try it

ok, good address at least

8:37 PM

I highly doubt it will work but you can try anyway

how do i know if its working

8:37 PM

when my u-boot image is not working

you should see a log

where

Try with factory U-Boot

8:38 PM

on serial

tux-linux

Try with factory U-Boot

you mean?

Download the latest update zip from Kobo, and take the U-Boot in upgrade/

8:40 PM

Try 0x17800000 maybe instead

8:41 PM

https://stackoverflow.com/questions/45631141/u-boot-start-addressand-u-boot-imx-file-contains According to this

U-boot start address?and u-boot.imx file contains?

I'm trying to understand embedded Linux principles and can't figure out addresses at u-boot output. How can i find the starting address of U-boot? For example, I have MX6-SabreSD board based on i...

8:47 PM

well, no

I guess the only option now would be to try the mw commands above Or just flash U-Boot to the MMC with skip=1

1 i need to be sure thats this is safe* 2 i could, but i dont think getting it to work will be easy and fast, and unplugging the sd card from the inside many many times is not a good idea. the sd port could be damaged, the battery plug after many plug-unplug cykles too, and just tinkering with the inside is risky, there are ribbons, voltage to short (edited)

Szybet

1 i need to be sure thats this is safe* 2 i could, but i dont think getting it to work will be easy and fast, and unplugging the sd card from the inside many many times is not a good idea. the sd port could be damaged, the battery plug after many plug-unplug cykles too, and just tinkering with the inside is risky, there are ribbons, voltage to short (edited)

Well, once you get a working U-Boot, you can boot the kernel or flash it with fastboot

8:53 PM

The goal is just to get it booting and have it loading the kernel at sector 81920 with 18432 load size

8:53 PM

Ooooh

8:53 PM

Maybe

8:53 PM

ums 0 mmc 0

8:53 PM

Does that work?

8:53 PM

Should export entire MMC over USB as a block device

tux-linux

The goal is just to get it booting and have it loading the kernel at sector 81920 with 18432 load size

and maybee adding some features to it to make usb to work

1

tux-linux

Does that work?

yes

8:55 PM

but

8:55 PM

whot now?

8:56 PM

if i overwrite the bootloader and its bircked i will need to open the device

it isin't possible?

9:19 PM

https://stackoverflow.com/questions/31248779/boot-u-boot-from-ram-using-u-boot

Boot U-boot from RAM using U-boot

I would like to boot an U-boot image from RAM using another U-boot. I loaded the U-boot image in the RAM using tftp tftp 0x90000000 u-boot.img and tried to run it with go 0x90000000 but nothing hap...

Welp, maybe not

Szybet

if i overwrite the bootloader and its bircked i will need to open the device

yes

9:20 PM

But once you have a working bootloader you won't need to open it ever again

Szybet

1 i need to be sure thats this is safe* 2 i could, but i dont think getting it to work will be easy and fast, and unplugging the sd card from the inside many many times is not a good idea. the sd port could be damaged, the battery plug after many plug-unplug cykles too, and just tinkering with the inside is risky, there are ribbons, voltage to short (edited)

i writed here

9:22 PM

i would like to avoid that...

9:22 PM

https://stackoverflow.com/questions/53058002/imx6-get-u-boot-to-temporarily-boot-another-u-boot

iMX6 get U-Boot to temporarily boot another U-Boot

First some background: We have the following setup in our iMX6-based embedded system. There are two U-Boot partitions and two system (Linux) partitions. Currently we use only the first U-Boot parti...

9:22 PM

maybe it is possible?

9:22 PM

bruh

sunxi boards have an U-Boot which loads another U-Boot, it's called U-Boot SPL. I don't know if it is possible on i.MX though...

couldn't i write a second u-boot somewhere on the sd card and boot it then?

no

i dont need the partition, i want to get the bootloader to work

The load address is hardcoded in the SoC

and i could't boot from uboot then?

Did you try to flash it with dd if=u-boot.bin of=/dev/mmcblk0 bs=1K seek=1 skip=1?

9:25 PM

When it is powered on, the SoC searches for something at sector 0x400 of the MMC, which is 1K. If it doesn't find anything there, it won't boot.

tux-linux

Did you try to flash it with dd if=u-boot.bin of=/dev/mmcblk0 bs=1K seek=1 skip=1?

i think yes

9:26 PM

thats not the problem

Ok, so compiler/source/other issue then

to resolve it i would prefer not to unplug the sd card every time

On all my Kobos, I have plugged/unplugged the SD card countless times (like, really a lot, 5-10 times/day for 6 months) and they still work

9:28 PM

I don't know how the Nia's slot is, but on the ones I have it is quite sturdy

shorting / a capacitor could fell off

9:28 PM

:/

It never happened to me either

good, but its a risk

9:29 PM

i dont like risk

And the number of times I've let my serial line, or my VCC line fall on the Kobo mainboard is quite high too, all works fine after that though

tux-linux

And the number of times I've let my serial line, or my VCC line fall on the Kobo mainboard is quite high too, all works fine after that though

its risk too

9:30 PM

hmm, now look at this

9:30 PM

Szybet

https://stackoverflow.com/questions/53058002/imx6-get-u-boot-to-temporarily-boot-another-u-boot

i used the dd there

You can take Clay Tack and create a small patch of it, then take out the SD with it

Szybet

Click to see attachment 🖼️

it hanged after it

9:31 PM

bruh

Szybet

i used the dd there

I guess it really didn't know how to handle that file, so it just reset instead of outputting data abort errors

tux-linux

I guess it really didn't know how to handle that file, so it just reset instead of outputting data abort errors

then there would be data error, or smth like that

9:32 PM

and this looks like it booted but then hang

9:32 PM

Maybe not

tux-linux

I guess it really didn't know how to handle that file, so it just reset instead of outputting data abort errors

We know this is the newer UBoot simply because the normal one we're using outputs an October date rather than a November one. Unfotunately, it hangs at that point, with the watchdog timer eventually kicking in and rebooting back to the original UBoot but I suspect that has to do with UBoot not liking the current state of the device (i.e., it doesn't like initialising it twice). So we'll have to figure out how to convince it to do so but at least we've gotten it booting another copy of itself, which is what the question was about.

Clay Tack really works well to attach/detach MMC

Szybet

We know this is the newer UBoot simply because the normal one we're using outputs an October date rather than a November one. Unfotunately, it hangs at that point, with the watchdog timer eventually kicking in and rebooting back to the original UBoot but I suspect that has to do with UBoot not liking the current state of the device (i.e., it doesn't like initialising it twice). So we'll have to figure out how to convince it to do so but at least we've gotten it booting another copy of itself, which is what the question was about.

oh no... ;/

Then download mode is really the last option

tux-linux

Clay Tack really works well to attach/detach MMC

well yes the sd card isin't that much of a problem

9:34 PM

but look at this

9:34 PM

this connector with a chip on it

9:34 PM

eassly to damage

9:35 PM

i would really really prefer not to open it every time

Maybe you can just leave the back over off?

9:35 PM

Once I open my Kobos they don't get closed before a long time ;p

tux-linux

Maybe you can just leave the back over off?

its risky to the connector

9:37 PM

i dont close it for now, just some rubber bands

9:38 PM

lets try from the other site, why does inkbox need to change those memory locations

9:38 PM

if its bigger, then why? could't it fit?

9:38 PM

maybe do some kernel options as modules, they are stored in the initframs

Because otherwise it overwrites the waveform

whots waveform in the kobo

= display completely stops working

Szybet

maybe do some kernel options as modules, they are stored in the initframs

and this idea?

What do you mean?

the kernel is too big, right?

yes

then compile more modules, the are stored in the initframs

~8MB on a Wi-Fi device

?

Szybet

then compile more modules, the are stored in the initframs

The thing is that the modules are embedded in the initramfs

9:41 PM

in a squashfs

thats the point

So that I don't have to mess up on the rootfs to setup the right modules for each devices

Szybet

thats the point

yes

9:41 PM

If I compile more modules, it will only make matters worse

i mean

9:42 PM

compile things in the kernel as modules

9:42 PM

the kernel will be smaller

no

9:42 PM

because the initramfs is embedded in the kernel

whot... how... why...

Because it is possible

if it wasn't like that it could work without modyfication

9:44 PM

i dont know if it must be like that

Maybe another option would be to load the kernel from a partition But I think the vendor U-Boot prevents that and loads it from the unpartitioned space anyway

9:45 PM

You can try

9:45 PM

extload command

9:45 PM

But at this point, there's got to be a way to compile U-Boot I mean, it's not possible that it can't work

9:46 PM

Did you try flashing Kobo's provided U-Boot to your MMC? Does that work?

tux-linux

Did you try flashing Kobo's provided U-Boot to your MMC? Does that work?

i think yes, but i will try it one more time to be sure

Be sure to flash this file upgrade/mx6ull-ntx/u-boot-mx6ull-E60U20-LPDDR2-256MB.imx

9:47 PM

dd if=u-boot-mx6ull-E60U20-LPDDR2-256MB.imx of=/dev/mmcblk0 bs=1K seek=1 skip=1

yes i use this

And it doesn't work?

tux-linux

And it doesn't work?

i think it worked, i will try it now

1

it does not work?

9:57 PM

thats... weird?

Yes it is

9:57 PM

So maybe what you compiled was good

no, i restored it to work by dd 30M of the sd image

9:58 PM

so the old is working

Yes, but if the provided official one does not work, there is a problem

but if i try to flash it kills himself

9:58 PM

https://github.com/Szybet/kobo-nia-audio/blob/main/files-from-device/init.d/upgrade-generic.sh

kobo-nia-audio/upgrade-generic.sh at main · Szybet/kobo-nia-audio

Kobo nia ereader with audio added. Contribute to Szybet/kobo-nia-audio development by creating an account on GitHub.

what--

here is literally the update script

tux-linux

what--

it just bricks

oh the device

9:59 PM

when you reboot it you mean?

yes

that's really strange

i will copy the bricked fragment and look at the hex

1

i dont know how to show this to you

10:15 PM

but in the working one there is some data before the u-boot

10:15 PM

and the not working one does not

Szybet

i dont know how to show this to you

click watch oryginal for better wiev

I don't really know what to say

10:16 PM

There seem to be differences between the two

10:16 PM

But I can't interpret them

there are maybe diffrent u-boot versions, but the concerning thing is the data that isin't in the bricked

10:18 PM

its maybe a header for the cpu, to check for something

10:19 PM

like, android devices kernels need special headers to boot

hmm... @andi so whot about the download mode?

tux-linux

If there is no download_mode command in U-Boot, then you could enter this:

eBR-1A # mw 20d8040 30
eBR-1A # mw 20d8044 10000000
eBR-1A # reset

But I suggest you wait a bit; @andi might have some warnings since it's a different platform

i mean this

im dumb

11:23 PM

i figured it out

11:23 PM

i can now write using dd

11:23 PM

the command is dd if=$UBOOT of=/dev/$DEVICE bs=1K seek=1 without skip

the custom kernel is working, i needed to use the .imx file

Szybet

im dumb

So what was the problem?

4:49 AM

Nice!

tux-linux

So what was the problem?

1 the command was wrong, without skip, not with it 2 idk, why when i tested without skip it erased the dos, maybe the sd card reader was faulty this one time 3 i needed to use the compiled .imx file, not been like you said

1

ok, now im waiting for instructions whot to change in the kernel and bootloader

1:40 PM

and how to install inkbox

generally abut uboot and skip: if the uboot.imx binary starts with zeros, you need the skip

3:50 PM

if not, you do not need it

? its working now

3:51 PM

it is working

3:51 PM

thats the point now

3:51 PM

maybe the bin file would work too

3:51 PM

idk

no

andi

generally abut uboot and skip: if the uboot.imx binary starts with zeros, you need the skip

good to notice

that download mode needs the uboot.imx without all that leading zeros (edited)

i dont have the command for download mode

3:55 PM

and i didn't saw a option for it in the config by menuconfig?

that stone-age 2009.08 kobo-patched uboot prepends the zeros in the build process, and fastboot flash bootloader strips them. Later uboots do not prepend that zeros.

1

3:55 PM

seems not to be there in newer uboots

okay

but you can do the two magic mw commands

are they safe?*** (edited)

i will check for the 6ull

hm? how?

I will check the data sheet of the SoC

1

oh, cool

for the 6sl and 6sll they are safe

3:57 PM

just writing to some volatile registers

3:57 PM

which get reset on some power cycle

@tux-linux you could describe the steps to install inkbox, I can now compile the kernel and u-boot

It's really more like doing it as we go, there isn't a premade manual or anything You can look at the Libra H2O porting thread to get an idea

hmm, okay

I'll guide you I'll be ready in some 10mins

sure i will wait

So you got a kernel compiled, and it works, right?

i have compiled many kernels, the work

4:10 PM

so yes

in the kobo you mean?

4:10 PM

because sometimes they are finicky about compiler used

for the kobo, yes

4:10 PM

i tryied to make usb host to work, so i compiled many of them

ok nice (edited)

4:11 PM

So what I'll first do will be to integrate the vendor sources into InkBox kernel tree

hmm okay

Did you modify anything to get it working, or it compiled and started fine out of the box?

look here

4:12 PM

https://github.com/Szybet/kobo-nia-audio/blob/main/links.md

kobo-nia-audio/links.md at main · Szybet/kobo-nia-audio

Kobo nia ereader with audio added. Contribute to Szybet/kobo-nia-audio development by creating an account on GitHub.

4:12 PM

Fixes first 2 points

4:13 PM

maybe your toolchains are also ok, idk

Does it work if you try to compile with armv7l-linux-musleabihf?

i didn't tryied it

4:13 PM

i can now if you want

yes, sometimes they compile fine but they don't boot

4:14 PM

so ideally try to start it on the Kobo too

message.txt

9.92 KB

4:27 PM

it does not work

You said it is Linux 4.1.15?

yes

Can you try with arm-nickel-linux-gnueabihf?

one sec

I know it is GCC 4.9, but it seems to compile fine my Libra's kernel which is also 4.1.15

4:35 PM

Also, does your Nia have any other physical buttons somewhere or do you only have the power button?

tux-linux

Also, does your Nia have any other physical buttons somewhere or do you only have the power button?

ony the power button

ok

4:37 PM

The Wi-Fi device uses 8189fs module I guess

4:38 PM

Do you have a picture of the PCB?

tux-linux

The Wi-Fi device uses 8189fs module I guess

something like this, look here to be sure https://github.com/Szybet/kobo-nia-audio/tree/main/logs

kobo-nia-audio/logs at main · Szybet/kobo-nia-audio

Kobo nia ereader with audio added. Contribute to Szybet/kobo-nia-audio development by creating an account on GitHub.

tux-linux

Do you have a picture of the PCB?

i have, i will upload them to github, one sec

4:39 PM

it compiled with nickel, i will flash it and try

Szybet

something like this, look here to be sure https://github.com/Szybet/kobo-nia-audio/tree/main/logs

it is, thanks the RTL871X gives it away

https://github.com/Szybet/kobo-nia-audio/tree/main/images/nia

kobo-nia-audio/images/nia at main · Szybet/kobo-nia-audio

Kobo nia ereader with audio added. Contribute to Szybet/kobo-nia-audio development by creating an account on GitHub.

4:46 PM

here are the images

4:46 PM

why do you need them?

well it was to verify the wi-fi chip but actually the dmesg logs were convincing enough

4:51 PM

But it will be useful anyway for future reference

4:52 PM

@andi will probably be interested to look at them

maybe andi1 could help me with usb host? :D

4:52 PM

actually who is andi1

Andreas Kemnade

4:53 PM

https://github.com/akemnade/linux

GitHub - akemnade/linux: Linux kernel source tree

Linux kernel source tree. Contribute to akemnade/linux development by creating an account on GitHub.

like, he has the datasheets of cpus

He ported Linux 5.16 to other Kobos

wow, really :O?

4:54 PM

It would be nice if nia would have the kernel 5.16

4:54 PM

i could do that, but i dont know how

4:54 PM

i dont know cpp or c, so im out propably

He's already made a test branch for Aura SE (vendor kernel initially at 3.0.35) which I didn't have the time to try yet... Hopefully today

@tux-linux nickel booted

5:00 PM

from where did you get this toolchain? why is it named nickel

It's from @pgaskin

5:01 PM

nice

tux-linux

It's from @pgaskin

and he is?

https://github.com/pgaskin/NickelTC

GitHub - pgaskin/NickelTC: A dockerized, deterministic, automated, ...

A dockerized, deterministic, automated, fixed, and fully-relocatable build of @NiLuJe's toolchain for Kobo eReaders. - GitHub - pgaskin/NickelTC: A dockerized, deterministic, automated, fix...

oh this

He created NickelMenu

wow :O

5:01 PM

nickelmenu is awesome, like how did he hacked into kobo program

Patch the binary

?

Well, instead of writing code, patching in binary data

??...

I don't know more, my knowledge of binary patching stops here

binary patching

5:03 PM

i hear it first time

Take a hex editor and replace bits

oh no

5:03 PM

no no

And you can change the behavior of a program I guess

yes but... its nearly impossible

5:03 PM

like, some variables ok

5:03 PM

but not an entire menu

Well, ask him, he will probably have more info than me on this

if he reads it, he can answer

5:04 PM

ok, the toolchain works, now whot

git pull the kernel repo

5:05 PM

Added Nia sources and scripts to make it work with InkBox build https://github.com/Kobo-InkBox/kernel/commit/069fc2fb3f7544d7b93bd07489af071961d8f958

5:05 PM

env GITDIR=/home/build/inkbox/kernel TOOLCHAINDIR=/home/build/inkbox/kernel/toolchain/arm-nickel-linux-gnueabihf/ THREADS=16 TARGET=arm-nickel-linux-gnueabihf scripts/build_kernel.sh n306 root

5:05 PM

Change env vars depending on where you cloned the repo

5:06 PM

I strongly suggest you simply put the repo in /home/build/inkbox/kernel

Szybet

and he is?

the entity replying to your message

:D

tux-linux

Well, instead of writing code, patching in binary data

naa, that's kobopatch

Oh sorry then But kobopatch must be related to NickelMenu in some way, right?

Szybet

nickelmenu is awesome, like how did he hacked into kobo program

he didn't (except for when he worked on the dictionary encryption in dictutil)

tux-linux

Oh sorry then But kobopatch must be related to NickelMenu in some way, right?

completely separate

Well, NickelHook does hack into the program

5:08 PM

hmm ok

tux-linux

Well, NickelHook does hack into the program

technically

pgaskin

he didn't (except for when he worked on the dictionary encryption in dictutil)

whot is this program doing

Is it like an LD_PRELOAD trick that hijacks the QPA and Nickel?

tux-linux

env GITDIR=/home/build/inkbox/kernel TOOLCHAINDIR=/home/build/inkbox/kernel/toolchain/arm-nickel-linux-gnueabihf/ THREADS=16 TARGET=arm-nickel-linux-gnueabihf scripts/build_kernel.sh n306 root

This should give you a small kernel that can still boot Nickel in kernel/out/n306/zImage-root Try it and report back if it boots

do i need to change n306 and root at the end to smth else?

no

5:11 PM

just dd if=zImage-root of=/dev/mmcblk0 bs=512 seek=2048

5:12 PM

I'll also grab U-Boot sources and put them into the tree as well

n306 is nia codename?

yes

@tux-linux how many changes were needed for the kernel aside from the oryginal? are there enabled chipidea controllers, and usb host support? is it the same kernel version?

For InkBox? Nothing, usually

5:16 PM

I only modify the kernel config

5:16 PM

not the code

i mean the kernel config

unless there's something really messed up

whot was there edited

oh; well namely, you have to enable loopback device support, FUSE, squashFS and other things

5:18 PM

i'll do that for you

5:18 PM

But once I set up the config, Nickel won't boot anymore since InkBox is a completely different infrastructure

tux-linux

But once I set up the config, Nickel won't boot anymore since InkBox is a completely different infrastructure

i dont care

5:18 PM

inkbox has usbutils by default, i think

5:18 PM

thats more important

5:20 PM

linux booted, kernel too

1

Szybet

inkbox has usbutils by default, i think

No?

whots this

5:28 PM

Alpine default

5:29 PM

Well maybe you can do something with it

5:29 PM

mx6ull_ntx_lpddr2_256m_defconfig so that U-Boot config worked?

tux-linux

Is it like an LD_PRELOAD trick that hijacks the QPA and Nickel?

yes

oh nice

tux-linux

mx6ull_ntx_lpddr2_256m_defconfig so that U-Boot config worked?

yes

tux-linux

Well maybe you can do something with it

well.... it will be easier to install libusb on your open source system that on kobo's default one (edited)

maybe not

5:32 PM

Have you looked on how InkBox OS root filesystem is mounted at boot?

5:32 PM

It's a read-only squashFS archive

5:32 PM

It can be made read-write with overlayfs-fuse though

5:33 PM

So actually yes, according to your recent enough kernel version you should be able to install things without issue

yea, and most imporant, kobo is making some weird udev parkour with usb, i tryied to disable it but with no success i think

You have to use the --no-cache option of apk to install packages though, otherwise it won't work

on your system i can see everything

1

tux-linux

You have to use the --no-cache option of apk to install packages though, otherwise it won't work

does it have a package manager???

yes, it's based on Alpine Linux

oh god thanks

5:34 PM

and oh god no

But it's an outdated version

5:34 PM

Alpine 3.10 from 2019

i installed alpine on my tablet and use it as a small server

5:34 PM

and installing software on it was pain

tux-linux

You have to use the --no-cache option of apk to install packages though, otherwise it won't work

why no cache?

If I used more recent versions of Alpine on devices with 2.6.35.3 kernels (>70% of supported devices), all (literally) binaries segfaulted and I couldn't get the system to work

Szybet

why no cache?

no idea

5:35 PM

It doesn't work if I omit it

weird

tux-linux

If I used more recent versions of Alpine on devices with 2.6.35.3 kernels (>70% of supported devices), all (literally) binaries segfaulted and I couldn't get the system to work

do you know why?

well, running a 2021 system on a 12 year old kernel isn't a good idea from the start

true

Interestingly enough, when you compiled latest musl and busybox yourself, it ran without issue

5:37 PM

So I tried to recompile all the Alpine packages on a Raspberry Pi but it ended up being waaaay too much work

also, why squashfs filesyste?

tux-linux

So I tried to recompile all the Alpine packages on a Raspberry Pi but it ended up being waaaay too much work

true, its too much effort. 2019 is good

1. Saves space 2. Read-only, if you set it to read-write you just delete the overlayfs cache to have a clean rootfs again 3. No MMC writes unless you have read-write enabled (well, read-only mount on standard partition works too, but squashfs is even better) (edited)

right

Also, it is oh-so-helpful for updates

the only downside is that is not so easy to install software on it

You just put a rootfs.squashfs file in the update package and scripts handle this for you

Szybet

the only downside is that is not so easy to install software on it

Well, overlayfs on newer devices like the Nia, Clara HD, Libra, makes this painless

5:40 PM

ifsctl mnt rootfs rw and you're good to go

whot about running somethink like this? https://github.com/baskerville/plato

GitHub - baskerville/plato: Document reader

Document reader. Contribute to baskerville/plato development by creating an account on GitHub.

5:41 PM

i think without issue?

https://github.com/Kobo-InkBox/rootfs/blob/master/etc/init.d/update.sh Update script

rootfs/update.sh at master · Kobo-InkBox/rootfs

InkBox OS root filesystem. Contribute to Kobo-InkBox/rootfs development by creating an account on GitHub.

Szybet

whot about running somethink like this? https://github.com/baskerville/plato

why would you want to do that?

i can program in rust, and its written in rust

5:41 PM

and its super fast

But InkBox already has a built-in ebook reader!

5:42 PM

I did not make InkBox so that people could install KOReader and Plato on it

if it will be realy good i will need to learn cpp!

tux-linux

I did not make InkBox so that people could install KOReader and Plato on it

thats true

Szybet

if it will be realy good i will need to learn cpp!

It's not that hard

for whot are we waiting for?

5:43 PM

to install inkbox on nia

tux-linux

It's not that hard

i programmed some arduinos, stm32 with cpp and it was sometimes pain, rust is easier

Szybet

to install inkbox on nia

compile u-boot

5:43 PM

will be there in a moment

5:43 PM

with modified code

tux-linux

compile u-boot

i need to do it?

5:44 PM

compile u-boot with nickel?

yes, to change kernel sector to 81920 and load size to 18432

5:45 PM

git pull and compile U-Boot in bootloader/mx6ull-n306/

tux-linux

yes, to change kernel sector to 81920 and load size to 18432

it isin't in the config, right?

https://github.com/Kobo-InkBox/kernel/commit/eee6be472914973e15dc6edbb7de4dd20ae0cc1f

N306: Change kernel load sector & size · Kobo-InkBox/kernel@eee6be4

5:49 PM

No, in ntx_comm.c

okay

5:50 PM

now i need to flash it?

yes

5:50 PM

And it should fail to boot

5:50 PM

But you should at least see a boot log

5:50 PM

Something saying that it can't find the kernel image

okay

Would be nice if you posted the boot log too

message.txt

2.52 KB

5:54 PM

here

5:54 PM

it booted, no linux kernel

Awesome

5:54 PM

Now, we'll partition the SD card

5:57 PM

Disk /dev/nbd0: 4 GiB, 4294967296 bytes, 8388608 sectors
Units: sectors of 1 * 512 = 512 bytes
Sector size (logical/physical): 512 bytes / 512 bytes
I/O size (minimum/optimal): 512 bytes / 512 bytes
Disklabel type: dos
Disk identifier: 0x25c915c4

Device      Boot   Start     End Sectors  Size Id Type
/dev/nbd0p1        49152   79871   30720   15M 83 Linux
/dev/nbd0p2       104448 1128447 1024000  500M 83 Linux
/dev/nbd0p3      1128448 1390591  262144  128M 83 Linux
/dev/nbd0p4      1390592 8388607 6998016  3.3G 83 Linux

(edited)

5:57 PM

Make it look like this (start/end sectors are very important) (edited)

5:57 PM

(the first partition starts at 49152, right?)

so i need to plug it into pc?

Well, yeah, ums 0 mmc 0 from U-Boot if you want

whot tool to use? fdisk?

yes

6:00 PM

You can clear all the partition table with o

tux-linux

You can clear all the partition table with o

?

in fdisk cmd

okay

6:01 PM

i havent used fdisk, always cfdisk, gparted and other

6:06 PM

whyyyy has it 15Mb

6:06 PM

It's the boot partition

6:07 PM

Only contains flags

6:07 PM

http://inkbox.ddns.net:40//wiki/index.php?title=Boot_flags

Boot flags

so its okay?

6:08 PM

your has 19 M mine 15

yes

weird

Well my bad

6:08 PM

I tried to edit the fdisk output but forgot to change the size

6:09 PM

Done

6:09 PM

Because the emulator's MMC first partition starts at 40959

6:09 PM

But I modified it so it could match your Nia's layout

oh okay

6:09 PM

just some confusion

yeah

my sd card is bigger, can i make the last partition to the end or is after ir some data?

yes

6:11 PM

you can extend it

14 gb for books

6:13 PM

i had 6,9 using kobo

6:13 PM

not only for books

tux-linux

not only for books

yes i know

all good

good, i will save it

6:15 PM

fdisk without sudo worked?

6:15 PM

it worked

6:15 PM

whot now?

Szybet

fdisk without sudo worked?

strange

6:15 PM

are you sure it did?

lsblk shows yes

hmm ok

unplugged, plugged back and its there

6:16 PM

so its ok

fine now, clone the rootfs repo

done

env GITDIR="${PWD}" ./release.sh

6:18 PM

Should create you a nice rootfs squashfs

6:18 PM

run it as root

6:18 PM

you can look at the script if you want to be sure it's safe

there isin't anything that could be harmfull

6:19 PM

done

6:19 PM

really fast

Ok, so now in parent directory you should have rootfs.squashfs

6:20 PM

right?

no...

what's the script output?

[root@ArchSzybet rootfs]# env GITDIR="${PWD}" ./release.sh
Parallel mksquashfs: Using 8 processors
Creating 4.0 filesystem on ../rootfs.squashfs, block size 1048576.
[============================================================================\] 6979/6979 100%

Exportable Squashfs 4.0 filesystem, gzip compressed, data block size 1048576
        compressed data, compressed metadata, compressed fragments,
        compressed xattrs, compressed ids
        duplicates are removed
Filesystem size 73621.37 Kbytes (71.90 Mbytes)
        34.22% of uncompressed filesystem size (215152.91 Kbytes)
Inode table size 74379 bytes (72.64 Kbytes)
        25.76% of uncompressed inode table size (288710 bytes)
Directory table size 79735 bytes (77.87 Kbytes)
        46.50% of uncompressed directory table size (171486 bytes)
Number of duplicate files found 666
Number of inodes 8427
Number of files 6899
Number of fragments 131
Number of symbolic links 904
Number of device nodes 0
Number of fifo nodes 0
Number of socket nodes 0
Number of directories 624
Number of ids (unique uids + gids) 2
Number of uids 2
        root (0)
        szybet (1002)
Number of gids 2
        root (0)
        szybet (1002)
Root filesystem has been compressed.

(edited)

This looks alright

6:22 PM

ls ../rootfs.squashfs?

tux-linux

ls ../rootfs.squashfs?

yes

6:22 PM

something is wrong with the script

tux-linux

Ok, so now in parent directory you should have rootfs.squashfs

oh, parent directory

Szybet

yes

is it there or not?

right

6:22 PM

yes it is (edited)

ok; I saw you have uid 1002 there; this will cause problems

6:23 PM

delete the whole repo and clone it as root this time

6:23 PM

so that the files will be owned by root

tux-linux

env GITDIR="${PWD}" ./release.sh

then redo that

okay

6:25 PM

maybe add to the script chown -R root:root * for the future?

maybe yeah

6:25 PM

I was thinking about fakeroot

6:25 PM

but didn't take the time to do it yet

maybe too

6:25 PM

done with root

compressed too?

Root filesystem has been compressed.

1

Now do this in an empty directory

openssl genrsa -out private.pem 2048
openssl rsa -in private.pem -out public.pem -outform PEM -pubout

Then tar cJvf ../keys-n306.tar.xz . And send me keys-n306.tar.xz via PM (edited)

6:27 PM

Never ever share the private key with anyone else

tux-linux

Now do this in an empty directory

openssl genrsa -out private.pem 2048
openssl rsa -in private.pem -out public.pem -outform PEM -pubout

Then tar cJvf ../keys-n306.tar.xz . And send me keys-n306.tar.xz via PM (edited)

thats the part i didn't understand

I put my trust in you This set of keys will be used to sign all packages in InkBox

okay, is it so important to sign them?

Yes, to ensure integrity

6:29 PM

Also InkBox OS won't boot if it can't validate the signatures

it would be easier to do like kobo does, with a md5sum file

tux-linux

Also InkBox OS won't boot if it can't validate the signatures

this could be turned off

there's an option to do that

6:30 PM

install a developer key

6:31 PM

https://github.com/Kobo-InkBox/kernel/blob/master/initrd/common/developer-key

kernel/developer-key at master · Kobo-InkBox/kernel

Kernel sources for InkBox OS. Contribute to Kobo-InkBox/kernel development by creating an account on GitHub.

does alpine require that?

But InkBox was designed for security

well, sure

So I won't change my position on the subject

good security is always good

1

6:31 PM

but on the other side its a ereader

Look at Kindles

6:32 PM

Newer ones have an even more strict security policy and implement HAB (verified boot in-SoC) to ensure kernel is signed correctly

its only becouse to ensure that DRM will work corectly

6:33 PM

it does not have much to do with security

6:33 PM

and why do you need those private keys? could't i do it myself?

Yes, too, but it is easier for some tasks (GUI bundle) Also if somehow you decide not to contribute at some point, I won't be able to sign the packages without you (edited)

Gui bundle?

6:36 PM

I dont understand the second point

6:36 PM

like sure, private keys are needed

6:37 PM

Will i build the filesystem with those keys, and then upload them to inkbox repo?

Only public key

6:37 PM

You'll be the only person with me that will know the private key

Szybet

Gui bundle?

update.isa file in update packages Contains qt, python and inkbox

6:38 PM

It isn't versioned yet, so it's local work currently

okay

Now you can sign the rootfs

6:39 PM

openssl dgst -sha256 -sign private.pem -out rootfs.squashfs.dgst rootfs.squashfs

6:39 PM

public keys are embedded in the initrd

6:39 PM

let me put the one you shared in there

tux-linux

openssl dgst -sha256 -sign private.pem -out rootfs.squashfs.dgst rootfs.squashfs

done

now, copy those files into partition 3 of sd card (edited)

when we speak about security, whot about encrypting files ( books ). they could be easly accesed when unplugging the sd card

6:42 PM

and the lock screen would not help

6:43 PM

some books are expensive, it would be nice if the thiev of the ereader could't get them

already done

really?

hehe

6:43 PM

yes

6:43 PM

Look at the features list

oh here i see

6:43 PM

Encrypted storage with EncFS

That was a wild ride

tux-linux

now, copy those files into partition 3 of sd card (edited)

you mean ootfs.squashfs.dgst rootfs.squashfs private.pem?

Probably the most complicated script I ever made https://github.com/Kobo-InkBox/rootfs/blob/master/etc/init.d/encfs

rootfs/encfs at master · Kobo-InkBox/rootfs

InkBox OS root filesystem. Contribute to Kobo-InkBox/rootfs development by creating an account on GitHub.

Szybet

you mean ootfs.squashfs.dgst rootfs.squashfs private.pem?

not private.pem only rootfs.squashfs*

so 2 files?

yes

tux-linux

Probably the most complicated script I ever made https://github.com/Kobo-InkBox/rootfs/blob/master/etc/init.d/encfs

this looks terrifyingly

I know right

6:45 PM

it works so i'm happy

does it impact performance?

no

6:45 PM

Well, creating the encrypted storage archive can take a lot of time

6:45 PM

5 mins on extreme cases

6:45 PM

But once it's done it is fast to mount

i did ums 0 mmc 0 and i try to mount the partition and i cant (edited)

6:46 PM

do i need to format it?

Szybet

this looks terrifyingly

This is the function in InkBox GUI https://github.com/Kobo-InkBox/inkbox/blob/master/encryptionmanager.cpp (edited)

Szybet

do i need to format it?

yes

also, why only one partition is showing in dolphin

tux-linux

yes

ext4?

mkfs.ext4 -O "^metadata_csum" /dev/mmcblk0p3

6:47 PM

do the same for p1, p2 and p4

Szybet

also, why only one partition is showing in dolphin

because the other ones aren't formatted

tux-linux

because the other ones aren't formatted

this one isin't too

And partition 1 is the only one that still starts at the same sector

tux-linux

mkfs.ext4 -O "^metadata_csum" /dev/mmcblk0p3

is it with the size 128 Mb?

yes

whot is in the 500 Mb partition stored?

After that: Clone the emu repo Grab overlaymount-rootfs.squashfs from sd/ directory, sign it the same way you did with rootfs.squashfs and put the two files in p3 as well

Szybet

whot is in the 500 Mb partition stored?

recoveryfs

tux-linux

recoveryfs

okay

https://youtu.be/KaBj7acHRHk?t=2010 See here, it's a serial log from the Touch C, the recovery partition is featured in the middle of the section

Nicolas Mailloux

InkBox OS 1.5

tux-linux

After that: Clone the emu repo Grab overlaymount-rootfs.squashfs from sd/ directory, sign it the same way you did with rootfs.squashfs and put the two files in p3 as well

where is the sd directory? in which repo?

emu repo

6:52 PM

https://github.com/Kobo-InkBox/emu

tux-linux

https://youtu.be/KaBj7acHRHk?t=2010 See here, it's a serial log from the Touch C, the recovery partition is featured in the middle of the section

okay, it takes some time to launch it

tux-linux

emu repo

wait, does it run on the kobo using the emulator?

Szybet

wait, does it run on the kobo using the emulator?

No

6:56 PM

It's just that the file I want you to put there is available in this repo

confusing

6:57 PM

whot is this file?

It's a helper which provides squashfuse, unionfs-fuse and overlayfs-fuse binaries

6:58 PM

Was too big to put in the initrd

"helper"?

Basically a mini Alpine chroot

oh okay (edited)

you can mount it if you want, to see

6:59 PM

mount overlaymount-rootfs.squashfs /mnt

tux-linux

mount overlaymount-rootfs.squashfs /mnt

already looked into it

1

Szybet

Click to see attachment 🖼️

so its done

alright

7:00 PM

I'll set up the kernel config and we can have a try with what you have inside currently

7:01 PM

Meanwhile

cat sd/user.sqsh.* > sd/user.sqsh
sudo unsquashfs -f -d [location-of-p4]/ sd/user.sqsh

(edited)

7:02 PM

It's nice, the Nia has the same PPI as the Glo and Aura SE, so supporting it will be easier since all the configs are there for devices of 1024x768 resolution

tux-linux

Meanwhile

cat sd/user.sqsh.* > sd/user.sqsh
sudo unsquashfs -f -d [location-of-p4]/ sd/user.sqsh

(edited)

cat?

Szybet

cat?

Assemble multiple files user.sqsh was too big for GitHub (edited)

weird way to do that

I know

7:02 PM

I didn't find another one

7:02 PM

But it works no problem

whot is p4?

userdata

7:03 PM

Also, X11/KoBox images as well as GUI bundle

7:03 PM

and other things

from where i get it

7:04 PM

i dont really understand

Szybet

from where i get it

In the emu repo?

7:05 PM

Same place where you found overlaymount-rootfs.squashfs

sudo unsquashfs -f -d sd/user.sqsh sd/user.sqsh

7:05 PM

this doesn't see, right

what

7:06 PM

sudo unsquashfs -f -d /run/media/szybet/wherever/is/your/p4/mounted sd/user.sqsh

partition 4?

yes

now i understand

its done

7:15 PM

but sync takes time now

normal

7:15 PM

it's like 3G of data

the sd card is 7Mb/s max

7:15 PM

thats big

yes

7:16 PM

I'm setting up the kernel config now

would you like to have the default one and modify it?

i already have it

from where?

in the kernel source

there was it?...

arch/arm/configs/imx_v7_kobo_defconfig

oh...

7:21 PM

i used the config from /proc/config.gz

oh; i need one thing

?

where were the dmesg logs?

7:22 PM

I have to get the cmdline

https://github.com/Szybet/kobo-nia-audio/tree/main/logs

kobo-nia-audio/logs at main · Szybet/kobo-nia-audio

Kobo nia ereader with audio added. Contribute to Szybet/kobo-nia-audio development by creating an account on GitHub.

tux-linux

I have to get the cmdline

hm? whots that

ok I have it

7:23 PM

console=ttymxc0,115200 rootwait rw no_console_suspend hwcfg_p=0x8ffffe00 hwcfg_sz=110 waveform_p=0x8ff71a00 waveform_sz=582529 ntxfw_p=0x8ff71400 ntxfw_sz=1034 mem=255M boot_port=1 rootfstype=ext4 root=/dev/mmcblk0p1 quiet

Args passed to the kernel at boot

ah those

7:27 PM

could you enable usb options? usb otg, chipidea controller

Well, not in InkBox default config But you can change it It's quite easy

okay, i will do it later

it seems already enabled btw

7:35 PM

Maybe you can look into USB_AUDIO

sounds good

7:35 PM

first i need it to register any usb device as host

7:36 PM

you can look at dmesg-usb at whot I have achieved already and whots the problem (edited)

Ok, it's ready

7:44 PM

pull the repo

tux-linux

env GITDIR=/home/build/inkbox/kernel TOOLCHAINDIR=/home/build/inkbox/kernel/toolchain/arm-nickel-linux-gnueabihf/ THREADS=16 TARGET=arm-nickel-linux-gnueabihf scripts/build_kernel.sh n306 root

And redo this

7:45 PM

Make sure this time that your kernel repo is in /home/build/inkbox/kernel

tux-linux

Make sure this time that your kernel repo is in /home/build/inkbox/kernel

must this be the exact same path?

7:50 PM

fakeroot is a good idea to implement in the future (edited)

7:51 PM

well yes it must

  AS      arch/arm/vfp/vfphw.o
  ./scripts/gen_initramfs_list.sh: Cannot open '/home/build/inkbox/kernel/initrd/busybox-1.32.1/_install-n306'
make[1]: *** [usr/Makefile:73: usr/initramfs_data.cpio.xz] Błąd 1
make: *** [Makefile:947: usr] Błąd 2
make: *** Oczekiwanie na niezakończone zadania....

7:51 PM

i will do a symlink

7:55 PM

done

8:00 PM

propably now flash it?

Szybet

must this be the exact same path?

yes

already compiled

Szybet

propably now flash it?

dd if=zImage-root of=/dev/mmcblk0 bs=512 seek=81920

done

Well, now try to boot

8:03 PM

Please post boot log

8:03 PM

There will be errors

8:04 PM

I just want to know if kernel/initrd works fine

message.txt

3.36 KB

8:04 PM

bootloop

shit

8:04 PM

welp

8:04 PM

we have a problem

i dont know whot you changed so i cant really help

8:05 PM

maybe i have done something wrong

no

8:06 PM

probably related to load size

hmm

8:07 PM

it is

8:07 PM

kernel weighs 9.5M right?

8:07 PM

yes

ok so it's load size

its bad that it could't be more

8:08 PM

new devices, more hardware features ad the kernel will be bigger

Delete kernel/initrd/n306/opt/firmware.sqsh

8:08 PM

And recompile the kernel

Szybet

its bad that it could't be more

It can be more

so why it is not working?

Max kernel size we could have would be 11.5M

Szybet

so why it is not working?

Because load size was set to 18432 = 9.4M

whot is load size

I want to try to strip down the kernel before increasing load size

8:10 PM

Number of sectors/bytes U-Boot reads from MMC to load kernel

okay

tux-linux

Number of sectors/bytes U-Boot reads from MMC to load kernel

oh

8:10 PM

and it cant be more than actual file size?

8:11 PM

done, flash it?

Szybet

and it cant be more than actual file size?

yes

Szybet

done, flash it?

yes

8:12 PM

what's its size?

9.4

wow that's tight

8:13 PM

let's see

it stops at starting kernel

8:16 PM

ntx_gpio_get_value(404) : error parameter ! null ptr !
ESDin=0,UPGKey=-1,PWRKey=0,USBin=0x1,BootESD=0,MenuKey=0
mmc read 0x8ffffc00 0x37ff 0x1

MMC read: dev # 0, block # 14335, count 1 ... 1 blocks read: OK
mmc read 0x8ff71a00 0x3800 0x472

MMC read: dev # 0, block # 14336, count 1138 ... 1138 blocks read: OK
mmc read 0x8ff71800 0x405 0x1

MMC read: dev # 0, block # 1029, count 1 ... 1 blocks read: OK
[WARNING] Binaries load sequence should Lo->Hi !
mmc read 0x8ff71400 0x406 0x3

MMC read: dev # 0, block # 1030, count 3 ... 3 blocks read: OK
Kernel RAM visiable size=255M->255M
ntx_gpio_init():R [ERROR] GPIO group number error (0)!!
ntx_gpio_init():B [ERROR] GPIO group number error (0)!!
hwcfg rootfstype : 2
hwcfg partition type : 2,bootmode=0
ntx_prebootm : cmd=setenv bootargs ${bootargs}  hwcfg_p=0x8ffffe00 hwcfg_sz=110 waveform_p=0x8ff71a00 waveform_sz=582529 ntxfw_p=0x8t
Kernel image @ 0x80800000 [ 0x000000 - 0x95e370 ]
## Flattened Device Tree blob at 83000000
   Booting using the fdt blob at 0x83000000
   Using Device Tree in place at 83000000, end 8300d274

Starting kernel ...

(edited)

8:16 PM

after a while its rebooting

8:16 PM

but it stops for some secons at starting kernel

hmm

tux-linux

yes

thats bad... changing the kernel will be pain

8:17 PM

on the stock firmware it wasn't like that?

yes, it can be more than the actual file size

8:18 PM

that's what I meant

oh

I forgot something

so just update u-boot?

tux-linux

I forgot something

whot?

GITDIR="${PWD}" scripts/make_devicenodes.sh

8:18 PM

in kernel repo

8:18 PM

then recompile kernel

8:18 PM

maybe will it work (edited)

mknod

8:20 PM

i first see something like this

yes

8:21 PM

It's for making device nodes in /dev

8:21 PM

Such as the console ttymxc0

8:21 PM

Without that node, the kernel might hang

8:21 PM

Which I think is what happened here

data abort

message.txt

2.99 KB

Ok, don't know what to say

8:30 PM

hang on

8:32 PM

@andi , if you have a clue, I'm all ears

maybe i will write the stock kernel to that location to see if it boots?

you'd need to recompile u-boot

why

to change load sector

8:33 PM

oh

8:33 PM

yeah you could try the stock kernel

8:33 PM

at sector 81920

that whot i said

8:33 PM

i will try it

sure

a buetiful kernel panink

8:35 PM

Starting kernel ...

/cpus/cpu@0 missing clock-frequency property
sy7636-pmic sy7636-pmic: request vcom gpio failed (-16)!
sy7636-pmic sy7636-pmic: request powerup gpio failed (-16)!
syscon-poweroff 20cc000.snvs:snvs-poweroff: pm_power_off already claimed 80021934 ntx_machine_poweroff
PMU: ricoh61x_battery_probe : version is RICOH61x_BATTERY_VERSION: 2014.02.21 V3.1.0.0-Solution1 2015/02/09
==== calc_soc_by_voltageMethod E60U22 ====
imx6ul-pinctrl 2290000.iomuxc-snvs: pin MX6ULL_PAD_SNVS_TAMPER3 already requested by 21f8000.i2c; cannot claim for gpio-keys
imx6ul-pinctrl 2290000.iomuxc-snvs: pin-5 (gpio-keys) status -22
imx6ul-pinctrl 2290000.iomuxc-snvs: could not request pin 5 (MX6ULL_PAD_SNVS_TAMPER3) from group gpio_snvs_keys_grp  on device 22900s
gpio-keys gpio-keys: Error applying setting, reverse things back
Kernel panic - not syncing: VFS: Unable to mount root fs on unknown-block(179,1)
---[ end Kernel panic - not syncing: VFS: Unable to mount root fs on unknown-block(179,1)
imx_epdc_v2_fb 228c000.epdc: Cleared LUT complete for init update

that's good

8:35 PM

it makes sense

so the kernel is wrong

i don't know why it doesn't work with inkbox

maybe your kernel config was wrong?

8:37 PM

compare your config without modyfication with this one that works in 100% https://github.com/Szybet/kobo-nia-audio/blob/main/kernels/self-compiled-oryginal/config-oryginal-kernel

kobo-nia-audio/config-oryginal-kernel at main · Szybet/kobo-nia-audio

Kobo nia ereader with audio added. Contribute to Szybet/kobo-nia-audio development by creating an account on GitHub.

I highly doubt it, never seen a wrong config that triggers a data abort in U-Boot

maybe the kernel has some wrong headers at beginning?

no idea

8:38 PM

i'll concentrate on the aura 2 for some moments, I'll try andi1's kernel

8:38 PM

will come back in a short time

okay, sure

tux-linux

will come back in a short time

does it mean today or in the nearest future?

today

8:40 PM

won't take long

8:40 PM

10mins

okay

what if you delete initrd/n306/opt/splash-rootfs.tar.xz?

one sec

9:00 PM

yes, thats progress

9:00 PM

```

message.txt

2.68 KB

yaaay

whot was the problem?

what's the size of your compiled kernel?

7.5

9:02 PM

M

What's the latest commit that shows up in git log?

change kernel config

9:04 PM

What's git status?

9:05 PM

it says its up to date

9:05 PM

in master

oh

?

9:06 PM

you forgot to push?

pull again and execute make_devicenodes again

done

9:08 PM

have in mind the repo has some files deleted that you said before

So now there are nodes in initrd/n306/dev?

Szybet

have in mind the repo has some files deleted that you said before

no problems, a git reset --hard will get them back

tux-linux

So now there are nodes in initrd/n306/dev?

tux-linux

no problems, a git reset --hard will get them back

so i do it?

no (not yet) (edited)

okay

Szybet

Click to see attachment 🖼️

Nice, now recompile the kernel

and flash it?

yes

message.txt

2.63 KB

9:15 PM

now i noticed that the screen refreshed, i dont know if it was done before using inkbox kernel

pull again

9:18 PM

recompile again

9:18 PM

it should work

9:18 PM

I added init argument to cmdline

Szybet

now i noticed that the screen refreshed, i dont know if it was done before using inkbox kernel

good

Starting kernel ...

[    0.004763] /cpus/cpu@0 missing clock-frequency property
[    0.423820] sy7636-pmic sy7636-pmic: request vcom gpio failed (-16)!
[    0.423864] sy7636-pmic sy7636-pmic: request powerup gpio failed (-16)!
[    5.534496] syscon-poweroff 20cc000.snvs:snvs-poweroff: pm_power_off already claimed 80021c08 ntx_machine_poweroff
[    5.545643] PMU: ricoh61x_battery_probe : version is RICOH61x_BATTERY_VERSION: 2014.02.21 V3.1.0.0-Solution1 2015/02/09
[    5.610660] ==== calc_soc_by_voltageMethod E60U22 ====
[    5.841222] imx6ul-pinctrl 2290000.iomuxc-snvs: pin MX6ULL_PAD_SNVS_TAMPER3 already requested by 21f8000.i2c; cannot claim for gps
[    5.856657] imx6ul-pinctrl 2290000.iomuxc-snvs: pin-5 (gpio-keys) status -22
[    5.863820] imx6ul-pinctrl 2290000.iomuxc-snvs: could not request pin 5 (MX6ULL_PAD_SNVS_TAMPER3) from group gpio_snvs_keys_grp  s
[    5.878155] gpio-keys gpio-keys: Error applying setting, reverse things back
mount: mounting /dev/mmcblk0p1 on /mnt failed: Invalid argument
umount: can't unmount /mnt: Invalid argument
Linux inkbox 4.1.15 #1 SMP PREEMPT Sat Feb 19 21:21:19 CET 2022 armv7l GNU/Linux
InkBox OS, kernel build 3NMth0qo
Copyright (C) 2021-2022 Nicolas Mailloux <nicolecrivain@gmail.com>

Checking filesystems ...

e2fsck 1.46.2 (28-Feb-2021)
Error reading block 114689 (Invalid argument).  Ignore error? yes

Force rewrite? yes

(edited)

9:25 PM

now im waiting for something to happen

9:26 PM

look at this specially:

mount: mounting /dev/mmcblk0p1 on /mnt failed: Invalid argument
umount: can't unmount /mnt: Invalid argument

(edited)

did you format p1?

one sec

9:29 PM

this 15 Mb one?

yes

9:30 PM

please tell me that it wasn't formatted

9:30 PM

otherwise...

9:30 PM

we have a really big problem

i did this now ._ .

9:30 PM

bruh

9:30 PM

soooo whots the problem

Are you able to mount it on your PC?

Szybet

i did this now ._ .

yes, becouse of this

9:33 PM

now whot

Try to restart

9:34 PM

The Kobo

message.txt

3.67 KB

YAY!

i see more errors then before

tux-linux

we have a really big problem

whots with that?

you did not format p2

Szybet

whots with that?

nothing

Szybet

i see more errors then before

doesn't matter

okay

9:38 PM

whotnow

Now try git reset --hard

9:39 PM

And recompile the full package

do the node thing one more time?

I don't think it will be needed

9:39 PM

But it can't hurt

if it doesn't work and leads to data abord, we'll try increasing the load size

it stopped at starting kernel, now i wait but thats propably it

git status?

9:50 PM

up to date

Remove quiet from CONFIG_CMDLINE in kernel/config/config-n306

9:51 PM

That will probably be somewhat helpful

i dont see anything after this change

10:01 PM

starting kernel and nothing more

ok

11:13 PM

Then very likely the load size

i think that too

Edit bootloader/mx6ull-n306/board/freescale/mx6ull_ntx/ntx_comm.c Change DEFAULT_KERNEL_LOAD_SZ to 20480 (edited)

這個檔案可被U-BOOT,開發板.c 板引入 um okay

tux-linux

Edit bootloader/mx6ull-n306/board/freescale/mx6ull_ntx/ntx_comm.c Change DEFAULT_KERNEL_LOAD_SZ to 20480 (edited)

there isin't somethink like that

11:19 PM

didn't find

hmm

11:20 PM

oh

11:20 PM

DEFAULT_LOAD_KERNEL_SZ

done

11:22 PM

how do i compile the bootloader?

11:23 PM

it was somewhere in my history

Well, same way you compiled it yesterday

oh, just make -j 6 CROSS_COMPILE=arm-nickel-linux-gnueabihf- ARCH=arm

11:23 PM

okay

11:30 PM

message.txt

38.34 KB

11:30 PM

looks like fs error

11:31 PM

idk man im dumb...

i dont think there is anything in recovery partition

11:32 PM

i dont remember putting anything there

11:32 PM

looks like the problem

invalid argument

awesome

11:36 PM

format p2

11:36 PM

copy overlaymount-rootfs.squashfs* files into it

11:39 PM

Also echo rooted | dd of=/dev/mmcblk0 bs=512 seek=79872

tux-linux

Also echo rooted | dd of=/dev/mmcblk0 bs=512 seek=79872

whots that

To allow the device to run a root kernel

whot does mean a root kernel

A kernel allowing SSH, root login, access to the device, etc.

11:43 PM

See http://inkbox.ddns.net:40/wiki/index.php?title=General_information#Standard_and_Rooted

General information

ah this

11:43 PM

okay

11:46 PM

and after 2s: NOOOOOOOOO IT CRASHED

11:46 PM

Security policy not enforced; root access permitted.                                           
[   21.852131] EXT4-fs (mmcblk0p3): mounted filesystem with ordered data mode. Opts: (null)    
Verifying root filesystem's digital signature ...                                              
Done.                                                                                          
[   26.770665] EXT4-fs (mmcblk0p1): mounted filesystem with ordered data mode. Opts: (null)    
[   26.832827] EXT4-fs (mmcblk0p4): mounted filesystem with ordered data mode. Opts: (null)    
FATAL: InkBox GUI root filesystem's signature is invalid!                                      
Aborting boot and powering off ...                                                             
The system is going down NOW!                                                                  
Sent SIGTERM to all processes

11:46 PM

:/

nice!

11:46 PM

Now, you have to sign gui_rootfs.isa in p4

11:47 PM

Important thing (as root):

unsquashfs -f -d g gui_rootfs.isa
cd g
echo "n306" > opt/device
rm ../gui_rootfs.isa
mksquashfs . ../gui_rootfs.isa -b 1048576 -comp xz -Xdict-size 100% -always-use-fragments
cd ..
rm -rf g

Then you can sign it openssl dgst -sha256 -sign private.pem -out gui_rootfs.isa.dgst gui_rootfs.isa (edited)

Szybet

and after 2s: NOOOOOOOOO IT CRASHED

Awesome!

tux-linux

Awesome!

error code 2

11:49 PM

where is gui rootfs?

well, the splash

it isin't in emu

p4

OH

it was in user.sqsh

Szybet

it isin't in emu

it isn't (typo correction ;p) (edited)

where did i unpack it...

you unpacked it to p4

oh right

It is [p4]/gui_rootfs.isa

god

11:51 PM

its 23:50 and i dont think straight

np

tux-linux

Important thing (as root):

unsquashfs -f -d g gui_rootfs.isa
cd g
echo "n306" > opt/device
rm ../gui_rootfs.isa
mksquashfs . ../gui_rootfs.isa -b 1048576 -comp xz -Xdict-size 100% -always-use-fragments
cd ..
rm -rf g

Then you can sign it openssl dgst -sha256 -sign private.pem -out gui_rootfs.isa.dgst gui_rootfs.isa (edited)

can i do those steps not on the sd card?

yes

11:52 PM

but you only unpack 40M worth of data

oh then

11:55 PM

done, reboot?

11:55 PM

and whot is opt/device

11:55 PM

a identifier?

yes

11:55 PM

to both

11:56 PM

it shouldn't work right away, there should still be errors but it'll probably be better

message.txt

45.13 KB

11:58 PM

so close...

11:59 PM

modules are missing

11:59 PM

i will be here for 20 more minutes, than i will go to sleep, tomorrow i should have some time but not so much that today (edited)

I am currently editing the scripts in rootfs to include the Nia

12:01 AM

There is hope!

there is

i'll make you the missing ui bundle

12:03 AM

then you could probably see inkbox gui working

12:03 AM

uploading right now

? i thinked that the problem are modules

no

12:04 AM

that's not essential

so whot failed?

ui bundle signature verification

12:05 AM

also the fact that the ui bundle is not there currently

12:06 AM

wget 23.163.0.39/bundles/private/update-n306.isa

um okay, why i can't sign it by myself?

12:06 AM

using the key i generated

I signed them using the keys you sent me

12:07 AM

So it will work

oh okay

12:07 AM

whot did you need to update?

12:07 AM

and whot should i do

oh

12:10 AM

Actually it's called update.isa but it isn't really an update

12:11 AM

cp update.isa [p4]/update/update.isa (edited)

its a sign

12:11 AM

i know

contains inkbox binary, qt libs and python

12:15 AM

don't forget to sync, then reboot and hopefully you should get an ui at some point

yea, i put it in the wrong directory

12:15 AM

its hard to think

12:15 AM

i always sync

12:16 AM

so the update is actually a update, it updates the files

well, not really but it doesn't really matter

 [ ok ]                                                                                        
/usr/bin/inkbox.sh: line 32: can't create /sys/class/leds/pmic_ledsb/brightness: nonexistent dy
 [ ok ]                                                                                        
                                                                                               
Welcome to InkBox OS 1.9                                                                       
Kernel 4.1.15 on armv7l (/dev/ttymxc0)                                                         
                                                                                               
inkbox login: [   64.246990] IRQ gpio=0, IN_TEN_SYS=0xd1,PWR_ON_TIMES=0x5b,int_sts=0x40,when ig
[   64.255851] PMU2: Watchdog timeout triggered. ===============                               
[   67.445394] PMU: ricoh61x_displayed_work Half-Clear CC, init_pswr(100), cc_delta(0)

nice

12:17 AM

so now look on the screen

i have no GUI but there is login

within some time you should see an ui appearing

oh, now it is

post a picture!!!

very nice

12:18 AM

So orientation will be to fix (edited)

12:19 AM

Can you touch somewhere? does it appear to react?

12:19 AM

Login as root

12:19 AM

password is root

12:19 AM

echo 3 > /sys/class/graphics/fb0/rotate killall inkbox-bin

i think the touch is not correct

env QT_QPA_PLATFORM=kobo:touchscreen_rotate=90:touchscreen_invert_x=auto:touchscreen_invert_y=auto:logicaldpitarget=0 chroot /kobo /mnt/onboard/.adds/inkbox/inkbox.sh

tux-linux

echo 3 > /sys/class/graphics/fb0/rotate killall inkbox-bin

because of that try it and the last command above will restart inkbox

tux-linux

env QT_QPA_PLATFORM=kobo:touchscreen_rotate=90:touchscreen_invert_x=auto:touchscreen_invert_y=auto:logicaldpitarget=0 chroot /kobo /mnt/onboard/.adds/inkbox/inkbox.sh

it does not launch

wait a bit? (edited)

12:21 AM

what's the error?

now it worked?

12:22 AM

i think wrong formatted from discord

so the touch is working?

it is

12:23 AM

its working

yay!!

but i think it flickers too much

is your power led still blinking?

to much refreshers

12:24 AM

yes it is

well inkbox is like that i guess?

12:24 AM

do you have a video so I could see?

wait

1

Szybet

yes it is

Does running /opt/bin/shutdown_led turns it off? (edited)

i entered library, snd there is sync in progress. for how long will it be like that

Szybet

i entered library, snd there is sync in progress. for how long will it be like that

You aren't connected to Wi-Fi so it should fail shortly

it does not (edited)

Otherwise killall inkbox-bin and relaunch that env command above

Szybet

it does not (edited)

the power light?

the sync

It does not fail?

and serial isin't working

12:27 AM

this is not good

12:27 AM

oh now

bad serial connection

12:28 AM

try rearranging the wires

yea, those wires are bad

tux-linux

Does running /opt/bin/shutdown_led turns it off? (edited)

no

hmm

12:29 AM

find /sys | grep led

i could try a video

12:30 AM

but it stopped responding in scrable

you can do it tomorrow otherwise

tux-linux

find /sys | grep led

I'd just need that output

oh now it is

12:31 AM

message.txt

15.85 KB

Ok; echo 1 > /sys/devices/platform/leds/leds/GLED; echo 0 > /sys/devices/platform/leds/leds/GLED

yes, it turned off

1

12:34 AM

its hard to click the file button in the paint app and i cant exit so i will post some drawings

12:35 AM

https://www.dropbox.com/s/mgoczd99ekv05yv/vid_20220220_003221.mp4?dl=0

VID_20220220_003221.mp4

Shared with Dropbox

12:35 AM

its slower than in plato, but yea its just a paint

12:36 AM

oh goodnes how it flickers when i entered maps

12:36 AM

thats not good

Szybet

oh goodnes how it flickers when i entered maps

It does that with other Kobos too

thats not good, and not usable

But I was talking about a video showing the refresh

Szybet

thats not good, and not usable

Well, light maps is not really the main purpose of InkBox

true

12:37 AM

how do i exit this app

you can't

whot XD?

But relaunch InkBox from cmd (edited)

12:38 AM

After killall lightmaps lightmaps-bin (edited)

done, but its stupid

I know

the touch is not working?

12:40 AM

after i relaunched

Is there a dmesg random kernel trace?

the dmesg is really long...

12:41 AM

message.txt

392.21 KB

Are you sure it doesn't work?

yes

12:42 AM

i forst launched inkbox and then killed maps, maybe becouse of this?

maybe

12:42 AM

kill inkbox-bin and relaunch it

12:43 AM

Maybe could you want to add QT_LOGGING_RULES=qt.qpa.input to env args

relaunched it, works

12:47 AM

idk why but the serial connection isin't great, for example when i paste the command only the half goes there

12:47 AM

and too much messages but that can be disabled

idk

12:47 AM

Nice that it works

12:47 AM

Almost finished updating my scripts

12:48 AM

Then I'll update cpp code for GUI and it will be fully ported! (edited)

12:48 AM

thanks for contributing

12:48 AM

come back tomorrow, we can test further some things

i thank you too

1

tux-linux

come back tomorrow, we can test further some things

i will, but i wont have that much time than today

1

12:49 AM

https://www.dropbox.com/s/fpc5cbbm5vo2kna/vid_20220220_004759.mp4?dl=0

VID_20220220_004759.mp4

Shared with Dropbox

12:50 AM

actually now it works good

12:50 AM

maybe the first launch was faulty

thanks

12:51 AM

if you want you can pull the rootfs repo

12:51 AM

updated the scripts

im going to sleep now

1

12:51 AM

i will do it tomorrow

it flashes alright

12:52 AM

no problem

now i have some time

@andi i have a question, its described here: https://github.com/kobolabs/Kobo-Reader/issues/116 in short, i think that the cpu handles usb support without any aditional chip, so could you check in the datasheet if it has support for host mode? not only gadget

[question] Kobo nia USB OTG as a host · Issue #116 · kobolabs/Kobo-...

First, I want to thank that this repository exists, I really appreciate that, It made kobo better than other ereaders on the market So here is my question, is it possible to use the kobo nia as a u...

In a random datasheet on the internet i see only usb otg and something about /dev/host

usb storage is not working

12:48 PM

but i didn't updated

and i dont know how to add books?

1:09 PM

i pasted them in gutenberg directory

1:09 PM

and nothing

mount [p4]/onboard /mnt

2:58 PM

Put your books in there

2:58 PM

It's normal that USB storage isn't working yet We didn't compile modules

you could just write the instructions for that, i will make it in the future

Basically make ARCH=arm CROSS_COMPILE=arm-nickel-linux-gnueabihf- modules

3:11 PM

Then make ARCH=arm CROSS_COMPILE=arm-nickel-linux-gnueabihf- INSTALL_MOD_PATH=/tmp/modules And make squashfs of this directory

3:13 PM

But I'll make you the archive It's located in the initrd, in opt/modules.sqsh

I checked quickly the pdf preview, its better than in plato and faster than in koreader but the sliders sometimes dont respond, and it makes it unusable

1

3:34 PM

the library didn't loaded correctly? idk

sometimes it's difficult to reach them

3:34 PM

I pushed the modules to the kernel repo

Szybet

I checked quickly the pdf preview, its better than in plato and faster than in koreader but the sliders sometimes dont respond, and it makes it unusable

Faster? Better? Wow, I thought my PDF engine was the worst

the file chooser is the worst part, its just not made for touchscreen, and especielly for eink

1

3:36 PM

koreader is written in lua, so yea. and plato uses gestures not silders, and those gestures didn't worked for me

tux-linux

I pushed the modules to the kernel repo

So now you can recompile it, mass storage won't work right away but you can try to see if usbnet works (when connecting your Kobo via USB to your computer, it should show up as an Ethernet device) echo true > [p1]/flags/USBNET_ENABLE (edited)

3:36 PM

Also, before doing that, pull the rootfs repo and make a rootfs.squashfs again, sign it and copy it into p3

3:37 PM

I modified the CPP code yesterday, so I'll share a GUI bundle at 23.163.0.39 shortly

i will try those modyfication next week, and if not in the next month

oh

i dont have time for this in the nearest time, school sucks (edited)

Well, I can sure wait, but maybe I won't be able to support it in 1.9 directly

3:45 PM

But anyways, thanks for your time and if you happen to have some this week, please try and report back! P.S.: Updated GUI bundle now, available at http://23.163.0.39/bundles/private/update-n306.isa (edited)

Szybet

the file chooser is the worst part, its just not made for touchscreen, and especielly for eink

I don't really have an alternative, but you can search books with the search button, in "Local storage"

tux-linux

I don't really have an alternative, but you can search books with the search button, in "Local storage"

If inkbox wants to be used by normal users than this must dissapear (edited)

Well, I mean, it isn't that bad

it is it is

In meantime when i dont have time to do anything, could you describe how Windows are working in inkbox? Is it xorg, is it fbink or combined?

What's "Windows"?

like the file chooser

5:18 PM

thats a window

Main GUI is Qt+FBInk, and KoBox is X11

the whot

Well, KoBox is a way to run X11 in a Kobo

oh okay

5:20 PM

so if i would develop an app, a seperate binary using fbink could it be launched as a window, like the file chooser?

No

5:30 PM

Would need to be a cli, gtk or Qt app

why only gtk or qt? so i assume just a xorg app?

Yeah

5:31 PM

Or Qt native

5:31 PM

As InkBox GUI does

I have written some gui apps before, in https://github.com/emilk/egui so i think i could replace the file chooser with something bigger, nicer and more touch friendly

GitHub - emilk/egui: egui: an easy-to-use immediate mode GUI in Rus...

egui: an easy-to-use immediate mode GUI in Rust that runs on both web and native - GitHub - emilk/egui: egui: an easy-to-use immediate mode GUI in Rust that runs on both web and native

The GUI is integrated in Qt and would be difficult to replace

5:33 PM

See reader.cpp

tux-linux

The GUI is integrated in Qt and would be difficult to replace

i think that you call the file chooser, and get a path from it right?

Yes

so calling a external script and getting a string path from stdout would be easy too?

I guess?

5:35 PM

But resources consuming too

not much, rust is good at performance

5:36 PM

only the disk reading and some ram would be bad

5:36 PM

plato is the fastest ereader software out there, and its written in rust (edited)

5:38 PM

https://youtu.be/KaBj7acHRHk

Nicolas Mailloux

InkBox OS 1.5

5:39 PM

you dont show here those maps that were glitching out on my device, are they even working on yours?

i looked up in the datasheet, the magic mw commands in the bootloader should do the same thing as on imx6sl or 6sll

so whot it does?

8:37 PM

can you share this datasheet? i want usb host get to work and the datasheets on the internet are bad

usb host: probably just debugfs...

8:37 PM

mw magic: it overwrites the default configuration to boot from

8:38 PM

for the next resets until power down

andi

usb host: probably just debugfs...

no, i get these logs after echo host > role https://github.com/Szybet/kobo-nia-audio/blob/main/logs/dmesg-usb

kobo-nia-audio/dmesg-usb at main · Szybet/kobo-nia-audio

Kobo nia ereader with audio added. Contribute to Szybet/kobo-nia-audio development by creating an account on GitHub.

and in that case it is some kind of manufacturing mode where you boot from usb

andi

mw magic: it overwrites the default configuration to boot from

oh nice

you send uboot over usb

8:39 PM

so for devices with internal eMMC, it is the way to test a new u-boot before flashing

8:40 PM

cat /dev/zero >/dev/mmcblk0 also lets you end up in that mode

cool to know, in that way even more ereaders could run inkbox

or removing the internatl sdcard

Szybet

no, i get these logs after echo host > role https://github.com/Szybet/kobo-nia-audio/blob/main/logs/dmesg-usb

could you look at it? I would appreciate it

hmm, did you apply external power supply?

i powered the external device, but not the kobo

at least on the devices I needed some y-shaped cable

i ordered some usb ports to make a cable to power both of them, but i dont think thats the case?

8:42 PM

y-shaped?

that works for me

does this power the kobo, and the device at the same time?

yes

maybe thats it? why would it need it to be charged?

there might be protection diodes letting stuff go to vbus, probably higher voltages at the data lines than on vbus might be a problem

thats a good explanation

8:46 PM

does your cable also have the OTG ID pin wired to ground?

no

8:47 PM

usual, that is the signal to provide power at vbus

8:48 PM

but if that does not work, I would leave it alone to not confuse other devices

andi

usual, that is the signal to provide power at vbus

and to detect to which mode to switch, host or gadget

but there you have the debugfs for

yes, but some devices, and ic make it automaticly

yes, correct, that is the way it should be

true

8:51 PM

thanks, i will try it

I am using a similar setup at my bicycle, it charges via hub dynamo my ebookreader

8:52 PM

and my bicycle is a usb gadget providing gps and some other sensor data

it actually works when its charging

9:37 PM

thanks

9:37 PM

:D

Szybet

you dont show here those maps that were glitching out on my device, are they even working on yours?

Yes they do

hmmm, good but why not on nia

No wifi yet?

that makes sense :D

tux-linux

I pushed the modules to the kernel repo

where are they?

5:20 PM

and do i need to use the openssl thing with modules too?

5:20 PM

i maybe have some time now to do it

Szybet

where are they?

In initrd/n306/opt/modules.sqsh

Szybet

and do i need to use the openssl thing with modules too?

No

6:21 PM

Just recompile the kernel

6:22 PM

It should work (edited)

tux-linux

Just recompile the kernel

why

Well, to have the new modules

6:22 PM

They are embedded in initrd which is embedded in the kernel

6:22 PM

They are mounted at boot at /modules

why the kernel that i use now cant be used

Because it has the old initrd

6:23 PM

Which modules are not for this device

okay

@tux-linux could i get privileges to pin messages in the chat?

9:26 PM

i want to make a guide and it will help to track the steps

Will do, Discord preferences seem rather convoluted

Well, i had some time, to less to install the modules but i started creating the guide: https://github.com/Szybet/kobo-nia-audio/blob/main/inkbox.md There are some points where @tux-linux could you please help? i writed there everything at the beginning to know mine point is to make inkbox more hacking friendly, not only for mine project but for everyone. Also discord is such a horrible place where knowledge gets pulled in into a black hole.

kobo-nia-audio/inkbox.md at main · Szybet/kobo-nia-audio

Kobo nia ereader with audio added. Contribute to Szybet/kobo-nia-audio development by creating an account on GitHub.

11:43 PM

There is no point to hack into kobo's software when there is inkbox, i think that could be the main "feature" of inkbox, not only open source, secure but also easy to understand and make your ereader like you would like with some tinkerink and developer skills (edited)

1

Thanks

Like, really, this is very useful I'll address your points shortly (somewhere this weekend, might I hope) and send a PR to your repo (edited)

oooooh nia audio :O

But I mean, Discord is not so bad for development. Since they introduced threads it's way easier to organize and sort things as we like so that they only focus on one subject.

But the knowledge is getting list in a long run

dill

oooooh nia audio :O

A magnet connected thin module with lithium polimer battery at the back, with jack and usb support

tux-linux

Thanks

Like, really, this is very useful I'll address your points shortly (somewhere this weekend, might I hope) and send a PR to your repo (edited)

Thanks!

1

Ok, done with it, what do you think? https://github.com/Kobo-InkBox/kobo-nia-audio/blob/main/inkbox.md

kobo-nia-audio/inkbox.md at main · Kobo-InkBox/kobo-nia-audio

Kobo nia ereader with audio added. Contribute to Kobo-InkBox/kobo-nia-audio development by creating an account on GitHub.

And im back

8:14 PM

why did you forked an entire repo just to edit one file... i will add some changes to my repo, with notes for you whot to add next

well. because that's the way it should be?

maybe make a pull request / push it to the repo then?

isn't that what I did?

oh, when you make a pull requests it forks the repo for you

8:16 PM

okay, my bad

1

i have added some points for you to clarify:

https://github.com/Szybet/kobo-nia-audio/blob/main/inkbox.md I will upgrade the guide with further instructions soon

kobo-nia-audio/inkbox.md at main · Szybet/kobo-nia-audio

Kobo nia ereader with audio added. Contribute to Szybet/kobo-nia-audio development by creating an account on GitHub.

8:51 PM

now i want to make my ereader to work

8:52 PM

You could focus more on the "From sources" points, becouse it would be hard for me to write them, to reverse enginere your work

tux-linux

wget 23.163.0.39/bundles/private/update-n306.isa

i will apply them

tux-linux

But I'll make you the archive It's located in the initrd, in opt/modules.sqsh

I pulled the repo, and pasted modules.sqsh into p3

9:08 PM

modules didn't work

9:09 PM

the gui is still rotated wrong

9:10 PM

i will try to recompile the kernel

now the brightness does not work:

message.txt

9.45 KB

Szybet

i will try to recompile the kernel

the same

9:24 PM

The library button after sync gives me an error, i see only one book that I choosed earlier using the file manager (edited)

9:28 PM

Why does a EPUB take like 1s to change a page... weird. And its REALLY hard to click the home button in option, while reading a book (edited)

9:29 PM

Thats all i can do, im waiting for your instructions on how to make it work. I will have some time so i will apply your patches faster @tux-linux (edited)

9:34 PM

another one... after i tryied to shutdown, it gives me this information in serial: [ 941.008667] imx_epdc_v2_fb 228c000.epdc: waiting for EPD power down ,current state=1...

i'm there

10:41 PM

https://github.com/Kobo-InkBox/kernel/commit/eee6be472914973e15dc6edbb7de4dd20ae0cc1f

N306: Change kernel load sector & size · Kobo-InkBox/kernel@eee6be4

10:42 PM

That's the commit related to kernel sector & load size in the bootloader

Okay, i will add it, thanks

and that one

10:44 PM

https://github.com/Kobo-InkBox/kernel/commit/98c6b9d9f11d52f496c693f3515dc2ad79588cdf

N306: Update scripts · Kobo-InkBox/kernel@98c6b9d

10:44 PM

Basically I just changed the load size from 18432 to 20480

Szybet

Why does a EPUB take like 1s to change a page... weird. And its REALLY hard to click the home button in option, while reading a book (edited)

Really? Usually it is pretty easy on my devices...

Szybet

The library button after sync gives me an error, i see only one book that I choosed earlier using the file manager (edited)

Did you connect to Wi-Fi? (edited)

10:46 PM

If you recompiled and flashed the kernel (edited)

tux-linux

Really? Usually it is pretty easy on my devices...

it clickes the info button

tux-linux

Did you connect to Wi-Fi? (edited)

nope, i dind't see a option to do that

oooh

10:47 PM

wait

tux-linux

If you recompiled and flashed the kernel (edited)

the modules doesn't work

what?

10:47 PM

define doesn't work

Szybet

Click to see attachment 🖼️

+ usb storage dosn't work too

ok, modules error then

10:49 PM

what does ls -shal /modules output?

Szybet

I pulled the repo, and pasted modules.sqsh into p3

You don't paste modules.sqsh in p3, it's embedded in the initrd, at opt/modules.sqsh But if you recompiled the kernel then idk Did you pull the repo before?

yes, i pulled

tux-linux

what does ls -shal /modules output?

one sec

10:55 PM

now the kernel does not boot...

message.txt

40.37 KB

10:55 PM

wtf?

With what command did you compile the kernel?

env GITDIR="${PWD}" TOOLCHAINDIR="${PWD}/toolchain/arm-nickel-linux-gnueabihf/" THREADS=$(($(nproc)*2)) TARGET=arm-nickel-linux-gnueabihf scripts/build_kernel.sh n306 root

10:57 PM

hm... do i need to do it manually?

Ok, seems alright

10:57 PM

No

10:57 PM

Is it the first time you boot the kernel since

Szybet

another one... after i tryied to shutdown, it gives me this information in serial: [ 941.008667] imx_epdc_v2_fb 228c000.epdc: waiting for EPD power down ,current state=1...

this happened?

look, its weird 1 flashed the kernel 2 booted from it, i described the problems above 3 poweroff 4 1 hour passes 5 its bricked

what if you reboot?

how?

Well, hold down the power buttton for some seconds

10:59 PM

Also, what's the kernel's size?

i have tryied it, but i will try it one more time

10:59 PM

Should be alright

look at this line:

[    7.824733] Modules linked in:                                                              
[    7.827893] CPU: 0 PID: 49 Comm: kworker/0:3 Not tainted 4.1.15 #1                          
[    7.834128] Hardware name: Freescale i.MX6 Ultralite (Device Tree)                          
[    7.840357] Workqueue: events epdc_firmware_work_func

11:01 PM

something with modules

No, Modules linked in: only displays the loaded modules. If there were some, it would say: Modules linked in: g_ether sdio_wifi_pwr 8189fs for example

then idk

So now it just says that no modules are loaded

11:02 PM

me too

11:02 PM

Can you boot Nickel? Do you have another SD card you could try Kobo factory image on?

okay i will try, but it will take time, like 1h to write the image

then wait

11:03 PM

Does your screen flash when it boots the kernel?

11:03 PM

Or it panics before?

tux-linux

Does your screen flash when it boots the kernel?

nope

11:03 PM

nope

Try flashing the Nickel kernel at sector 81920 And see if it flashes It will panic (won't be able to find the rootfs) but it should flash iirc (edited)

Starting kernel ...                                                                            
                                                                                               
/cpus/cpu@0 missing clock-frequency property                                                   
sy7636-pmic sy7636-pmic: request vcom gpio failed (-16)!                                       
sy7636-pmic sy7636-pmic: request powerup gpio failed (-16)!                                    
syscon-poweroff 20cc000.snvs:snvs-poweroff: pm_power_off already claimed 800217f0 ntx_machine_f
PMU: ricoh61x_battery_probe : version is RICOH61x_BATTERY_VERSION: 2014.02.21 V3.1.0.0-Solutio9
==== calc_soc_by_voltageMethod E60U22 ====                                                     
imx6ul-pinctrl 2290000.iomuxc-snvs: pin MX6ULL_PAD_SNVS_TAMPER3 already requested by 21f8000.is
imx6ul-pinctrl 2290000.iomuxc-snvs: pin-5 (gpio-keys) status -22                               
imx6ul-pinctrl 2290000.iomuxc-snvs: could not request pin 5 (MX6ULL_PAD_SNVS_TAMPER3) from gros
gpio-keys gpio-keys: Error applying setting, reverse things back                               
Kernel panic - not syncing: No working init found.  Try passing init= option to kernel. See Li.
---[ end Kernel panic - not syncing: No working init found.  Try passing init= option to kerne.

11:07 PM

nickel boots ok

so screen flashes?

no

Can you try reflashing InkBox kernel? Maybe something in the MMC was overwritten for whatever reason and it caused the error you posted before... (edited)

what is at 2048? (edited)

11:15 PM

sector

11:15 PM

in inkbox

81920

11:15 PM

nothing, InkBox uses sector 81920 for kernel

there is nothing, really nothing?

well, probably remnants of old nickel kernels I guess?

hm, okay

but nothing is loaded from there at boot

after rewriting its the same

Welcome to the Mysterious and Incredibly Painful to Resolve Issues World™️! And if you remove kernel/initrd/n306/opt/modules.sqsh, then recompile it, what happens?

one sec

1

11:20 PM

could it be possible to skip the modules, like integrate them into the kernel ( i mean in menuconfig * not m )

No, because some can't always be loaded

11:21 PM

e.g., wi-fi modules

11:21 PM

otherwise battery would drain very fast

:/

11:23 PM

message.txt

10.84 KB

11:23 PM

not good

whatever that is

11:24 PM

@andi , do you have some insight on this? I'm lost

11:24 PM

hm

11:25 PM

Did you, at some point, write InkBox kernel to sector 2048 of the MMC?

11:25 PM

And there it could have overwritten some of the waveform, dtb, etc.

tux-linux

Did you, at some point, write InkBox kernel to sector 2048 of the MMC?

its possible, but you said there is nothing

11:25 PM

:D

Well, after ~6MB of sector 2048, there are DTB, ntxfw, waveform and other blobs that allow the kernel to boot (that's why the kernel needs to be moved to sector 81920, otherwise there would be problems (edited)

so now we know

11:26 PM

i will write 32 Mb of the old nickel image, and then write the inkbox bootloader

11:27 PM

is that all that i need?

hmm

11:27 PM

Be sure to save the partition layout

11:27 PM

Writing the old nickel image on top of that will work, but it will overwrite the partition layout

11:28 PM

However, if you go into fdisk after that and re-partition the MMC as it was before, you'll be able to mount partitions again and nothing will be lost (don't format them though, just re-partition the MMC) (edited)

where is the partition layout

11:28 PM

oh, like that

Well, you can do fdisk -l

i will make a copy of the whole sd card, and then do this

1

meantime you could look at the other problems, errors

Well, if porting InkBox to the Nia leads to a hardware brick, better to solve this first

11:39 PM

(I don't say it's an hardware brick, but it might be for some obscure reason that I don't know about)

this would be sad

11:44 PM

but whot module could brick it

11:46 PM

if you think this could be a brick, i will simply flash the entire nickel image

11:47 PM

also, can I dd the whole disk using ums from u-boot?

yeah, but I don't think ums is that stable for such a big write

11:47 PM

I'd just take the SD out

12:54 AM

nickel booted

12:54 AM

its not bricked

ok very nice

12:55 AM

Maybe MMC corruption then?

tommorow i will restore inkbox, and the first 30Mb sector + flash the inkbox bootloader and we will see

1

tux-linux

Maybe MMC corruption then?

idk, tommorow we will see

1

inkbox boots and shows error code 3

11:05 AM

there are all partitions exept p1

11:05 AM

and i cant create it

Wybór (domyślnie e): p

Wybrano partycję 4
Pierwszy sektor (2048-31116287, domyślnie 2048): 49152

Sector 49152 is already allocated.

11:06 AM

hmmm

11:06 AM

i will just clear the parition table

kernel without modules booted:

message.txt

11.54 KB

11:27 AM

kernel with modules booted, but here are the errors again:

 [ ok ]                                                                                        
 * WARNING: clock skew detected!                                                               
insmod: ERROR: could not load module /modules/arcotg_udc.ko: No such file or directory         
 [ !! ]                                                                                        
 * ERROR: modules_init failed to start                                                         
 * Mounting onboard storage filesystem ...                                                     
 [ ok ]                                                                                        
 * Splash screen ...                                                                           
 [ ok ]                                                                                        
 * Starting USBNet ...                                                                         
 * USB networking is disabled. Please set the USBNET_ENABLE flag and try again.                
 [ ok ]                                                                                        
 * WARNING: clock skew detected!                                                               
insmod: ERROR: could not load module /modules/arcotg_udc.ko: No such file or directory         
 [ !! ]                                                                                        
 * ERROR: modules_init failed to start

11:28 AM

here is the command that you requested:

kobo:~# ls -shal /modules                                                                      
total 1K                                                                                       
     0 drwxr-xr-x    5 user     user          52 Feb 20  2022 .                                
     1 drwxr-xr-x   29 root     root        1.0K Jan  1 00:16 ..                               
     0 drwxr-xr-x    4 user     user          37 Feb 20  2022 drivers                          
     0 drwxr-xr-x    3 user     user          31 Feb 20  2022 fs                               
     0 drwxr-xr-x    2 user     user          32 Feb 20  2022 wifi                             
kobo:~#

tux-linux

echo 3 > /sys/class/graphics/fb0/rotate killall inkbox-bin

i still need to do this :/

11:32 AM

usb storage does not work

11:32 AM

i dont know where is the wifi menu...

11:33 AM

but brightness works back, thats good

11:33 AM

system shutdown does not work

11:33 AM

so we are in the start, everything works exept modules @tux-linux

Did you update the rootfs?

how exactly?

I added the Nia in all scripts

you mean making a squash and replacing the one on the ereader? (edited)

Yes

2:58 PM

And pulling the repo

the rootfs is in the emu repo?

Kobo Nia support was added in this commit https://github.com/Kobo-InkBox/rootfs/commit/c9e39e40161ed82a616abf46686ec8edc2e4a8c5

Kobo Nia (N306) support · Kobo-InkBox/rootfs@c9e39e4

2:59 PM

In the rootfs repo

3:00 PM

env GITDIR="${PWD}" ./release.sh Does that remind you of something?

this compresses the rootfs

Szybet

[root@ArchSzybet rootfs]# env GITDIR="${PWD}" ./release.sh
Parallel mksquashfs: Using 8 processors
Creating 4.0 filesystem on ../rootfs.squashfs, block size 1048576.
[============================================================================\] 6979/6979 100%

Exportable Squashfs 4.0 filesystem, gzip compressed, data block size 1048576
        compressed data, compressed metadata, compressed fragments,
        compressed xattrs, compressed ids
        duplicates are removed
Filesystem size 73621.37 Kbytes (71.90 Mbytes)
        34.22% of uncompressed filesystem size (215152.91 Kbytes)
Inode table size 74379 bytes (72.64 Kbytes)
        25.76% of uncompressed inode table size (288710 bytes)
Directory table size 79735 bytes (77.87 Kbytes)
        46.50% of uncompressed directory table size (171486 bytes)
Number of duplicate files found 666
Number of inodes 8427
Number of files 6899
Number of fragments 131
Number of symbolic links 904
Number of device nodes 0
Number of fifo nodes 0
Number of socket nodes 0
Number of directories 624
Number of ids (unique uids + gids) 2
Number of uids 2
        root (0)
        szybet (1002)
Number of gids 2
        root (0)
        szybet (1002)
Root filesystem has been compressed.

(edited)

.

tux-linux

openssl dgst -sha256 -sign private.pem -out rootfs.squashfs.dgst rootfs.squashfs

Sign it there And copy the two files in p3

yea sure

3:01 PM

will this fix the modules?

Yep

3:02 PM

Well it should

3:03 PM

I'll make you another GUI bundle too I added quite a lot of things to InkBox since two weeks ago

the most important thing is the screen rotating

It will be fixed

3:06 PM

https://github.com/Kobo-InkBox/gui-bundle FYI, you can now generate the bundle yourself, but it requires you to compile InkBox Qt code before, which is painful to setup, so probably better if I just send it to you

GitHub - Kobo-InkBox/gui-bundle: InkBox GUI bundle contents and gen...

InkBox GUI bundle contents and generation script. Contribute to Kobo-InkBox/gui-bundle development by creating an account on GitHub.

qt is pain to setup..

Yes

3:08 PM

But it's not impossible

its still pain

3:14 PM

wow, everything works

3:14 PM

wifi is there

3:14 PM

brightness too

3:15 PM

well usb storage does not work

Very nice Let's see USB

3:15 PM

What happens if you do lsmod?

Szybet

wifi is there

It connects, really?

tux-linux

It connects, really?

haven't tried yet, but wifi networks are showing up

It seems to be somewhat difficult to connect to some networks in InkBox for some reason

3:17 PM

(There it should use dhcpcd I think)

3:17 PM

hmmm

Can you see something on your computer?

3:17 PM

dmesg | tail?

tux-linux

It seems to be somewhat difficult to connect to some networks in InkBox for some reason

isin't it just wpa-supplicant?

and dhcpcd

3:19 PM

wpa_supplicant is used to actually connect to the network, and then dhcpcd is launched to obtain a DHCP lease

3:19 PM

See scripts in https://github.com/Kobo-InkBox/rootfs/tree/master/usr/local/bin/wifi

on the pc in dmesg is nothing

3:21 PM

here is from inkbox:

message.txt

28.3 KB

3:21 PM

just usb plugged 1

And lsmod?

kobo:~# lsmod Module Size Used by 8189fs 1294344 0 sdio_wifi_pwr 593 0

strange

3:21 PM

find /modules

message.txt

2.55 KB

3:23 PM

there is usb gadget

ok, seems I didn't update the script for mass storage

3:24 PM

insmod /modules/drivers/usb/gadget/legacy/g_mass_storage.ko? (edited)

3:25 PM

Will probably lead to an error

connection failed with wifi

I'll test on my Libra which has the same kernel

Szybet

connection failed with wifi

Now try /usr/local/bin/wifi/connect_to_network.sh ssid_there passphrase_there

3:26 PM

And post output

[root@ArchSzybet resources]# insmod /modules/drivers/usb/gadget/legacy/g_mass_storage.ko
insmod: ERROR: could not load module /modules/drivers/usb/gadget/legacy/g_mass_storage.ko: No such file or directory

huh?

3:26 PM

I meant, on the Kobo

3:26 PM

Not on your host

the keyboard is horrible, or the touch is not accurate

tux-linux

huh?

wrong terminal...

Szybet

the keyboard is horrible, or the touch is not accurate

There could be a solution, we'll try it (newer platform plugin) Though it comes with a cost, we need to change more files and we have to rely on VNC graphics in X11 instead of directly printing damage events on the framebuffer with fbink_xdamage

kobo:~# insmod /modules/drivers/usb/gadget/legacy/g_mass_storage.ko                            
insmod: ERROR: could not insert module /modules/drivers/usb/gadget/legacy/g_mass_storage.ko: Ue
kobo:~#

Otherwise it displays very strangely, don't know why

Szybet

kobo:~# insmod /modules/drivers/usb/gadget/legacy/g_mass_storage.ko                            
insmod: ERROR: could not insert module /modules/drivers/usb/gadget/legacy/g_mass_storage.ko: Ue
kobo:~#

What's the error? Can't read the rest (edited)

kobo:~# insmod /modules/drivers/usb/gadget/legacy/g_mass_storage.ko
insmod: ERROR: could not insert module /modules/drivers/usb/gadget/legacy/g_mass_storage.ko: Unknown symbol in module
kobo:~#

3:31 PM

kobo:~# /usr/local/bin/wifi/connect_to_network.sh hotspotttssss 12345678
ifconfig: ioctl 0x8913 failed: No such device
Successfully initialized wpa_supplicant
rfkill: Cannot get wiphy information
ioctl[SIOCSIWAP]: Operation not permitted
no such user dhcpcd
dhcpcd-9.4.0 starting
eth0: waiting for carrier
timed out
dhcpcd exited
dhcpcd_fork_cb: truncated read 0 (expected 4)
DHCP request failed.

Ok, we'll try with udhcpc then

3:31 PM

udhcpc -i eth0?

Szybet

kobo:~# insmod /modules/drivers/usb/gadget/legacy/g_mass_storage.ko                            
insmod: ERROR: could not insert module /modules/drivers/usb/gadget/legacy/g_mass_storage.ko: Ue
kobo:~#

I'll get you back with module deps after testing on my Libra, hang on

tux-linux

There could be a solution, we'll try it (newer platform plugin) Though it comes with a cost, we need to change more files and we have to rely on VNC graphics in X11 instead of directly printing damage events on the framebuffer with fbink_xdamage

i dont understand, but in nickel the touchscreen was just a /dev/input/mouse1 device

3:32 PM

this worked:

[root@ArchSzybet resources]# ssh root@10.42.0.28
The authenticity of host '10.42.0.28 (10.42.0.28)' can't be established.
ED25519 key fingerprint is SHA256:58MR3tZxGQMSKnyf5y6N6F2scXOG2wQ7IC33dMsJmDA.
This key is not known by any other names
Are you sure you want to continue connecting (yes/no/[fingerprint])? yes
Warning: Permanently added '10.42.0.28' (ED25519) to the list of known hosts.
root@10.42.0.28's password: 
Welcome to InkBox!
* Warning *
Root filesystem is mounted read-only.
Invoke `ifsctl mnt rootfs rw' to make it read-write.

kobo:~# udhcpc -i eth0
udhcpc: started, v1.33.0
udhcpc: sending discover
udhcpc: sending select for 10.42.0.28
udhcpc: lease of 10.42.0.28 obtained, lease time 3600
kobo:~#

ok so that works

3:33 PM

I'll change it in the script

those damn small sliders

3:33 PM

i simply cant use them

3:34 PM

in brightness, wifi select, pfg reader

3:34 PM

why cant those be bigger :/

Maybe they could be It has to do with padding and DPI On older devices, they are monstrous because of the low screen resolution

3:35 PM

BTW, you can take screenshots with chroot /kobo fbgrab /tmp/dump.png

wait, maybe ui scalling

tux-linux

BTW, you can take screenshots with chroot /kobo fbgrab /tmp/dump.png

thats cool

What's cat /data/config/09-dpi/config?

tux-linux

What's cat /data/config/09-dpi/config?

187

whoops

3:36 PM

hang on

ui scalling factor has a slider to set it, and its hard too

echo 195 > /data/config/09-dpi/config

3:36 PM

Then rc-service inkbox_gui restart

3:37 PM

^ should take ~1m to launch

UI scalling does not save

3:37 PM

the arrows to change the settings site are hard to click too

tux-linux

echo 195 > /data/config/09-dpi/config

still the same

Maybe are you on a build where I didn't actually support the Nia

3:39 PM

cat /data/config/09-dpi/config?

187 still

3:39 PM

weird

echo 195 > /data/config/09-dpi/config killall inkbox-bin; env QT_QPA_PLATFORM=kobo chroot /kobo /mnt/onboard/.adds/inkbox/inkbox.sh (edited)

kobo:~# echo 195 > /data/config/09-dpi/config
kobo:~# 
kobo:~# killall inkbox-bin; chroot /kobo /mnt/onboard/.adds/inkbox/inkbox.sh
Terminated
iniparser: cannot open /mnt/onboard/.kobo/nightmode.ini
cat: can't open '/mnt/onboard/.kobo/Kobo/Kobo eReader.conf': No such file or directory
qt.qpa.plugin: Could not find the Qt platform plugin "xcb" in ""
This application failed to start because no Qt platform plugin could be initialized. Reinstalling the application may fix this probl.

Available platform plugins are: kobo, vnc.

Aborted
kobo:~#

oh

3:41 PM

edited command above

well it worked

3:43 PM

its better

ok, probably wrong build then

3:43 PM

It will last that way until you re-open Settings I guess

but still i need to click hard to choose the 1 on the keyboard

strange

just maybe make a bigger keyboard

3:44 PM

or the ui scalling could work for it

hmm

3:44 PM

Have a screenshot?

of whot

3:44 PM

keyboard?

yeah

3:49 PM

making a photo is faster

tux-linux

BTW, you can take screenshots with chroot /kobo fbgrab /tmp/dump.png

Don't you think that would be easier ;p?

3:49 PM

ah

sending the image to pc is harder

scp?

usb storage doesn't work, wifi didn't wanted to connect

kobo:~# udhcpc -i eth0
udhcpc: started, v1.33.0
udhcpc: sending discover
udhcpc: sending discover
udhcpc: sending discover
udhcpc: sending discover
udhcpc: sending discover
udhcpc failed to get a DHCP lease
udhcpc: sending discover
udhcpc: sending discover
udhcpc: sending discover
udhcpc: sending discover
udhcpc: sending discover
udhcpc failed to get a DHCP lease
^C
kobo:~#

ah

3:51 PM

echo true > /boot/flags/USBNET_ENABLE

3:51 PM

reboot

3:51 PM

then the Kobo will show up as an Ethernet device when plugged in via USB to your computer

3:51 PM

And you can do ifconfig usb0 192.168.2.2, then ssh root@192.168.2.2 to connect

okay okay

3:52 PM

i will know for the future

1

Szybet

Click to see attachment 🖼️

Tried on my Glo which has the same screen resolution, no issues at all clicking "1"

3:53 PM

At least I did document this (usbnet): https://inkbox.ddns.net/wiki/index.php?title=Boot_flags (edited)

Boot flags

well the keyboard is acceptable when the ereader is not placed horizontally on the table (edited)

3:58 PM

so skip that problem for now

Szybet

well the keyboard is acceptable when the ereader is not placed horizontally on the table (edited)

Probably touch input handler hardware issue

maybe

4:00 PM

but its usable, needs some practice to use well

4:01 PM

what's next? what do we need to make nia fully usable? usb storage, wifi and?

insmod /modules/drivers/usb/gadget/function/usb_f_mass_storage.ko insmod /modules/drivers/usb/gadget/legacy/g_mass_storage

4:01 PM

Does that work?

Szybet

what's next? what do we need to make nia fully usable? usb storage, wifi and?

it's pretty much it ;D

4:02 PM

Then I'll have to make the recoveryfs, but that will happen just before the release

the brightness slider works great when the dpi is set to 195

1

4:04 PM

sliders in pdf reader are acceptable

1

tux-linux

insmod /modules/drivers/usb/gadget/function/usb_f_mass_storage.ko insmod /modules/drivers/usb/gadget/legacy/g_mass_storage

?

kobo:~# insmod /modules/drivers/usb/gadget/function/usb_f_mass_storage.ko
insmod: ERROR: could not insert module /modules/drivers/usb/gadget/function/usb_f_mass_storage.ko: Unknown symbol in module
kobo:~# insmod /modules/drivers/usb/gadget/legacy/g_mass_storage
insmod: ERROR: could not load module /modules/drivers/usb/gadget/legacy/g_mass_storage: No such file or directory
kobo:~#

ok, it's missing some deps

do other kobo ereader have audio support ( i mean integrated jacks )?

no

so if i want add audio support to inkbox i need to fork the entire project right?

insmod "/modules/fs/configfs/configfs.ko"
insmod "/modules/drivers/usb/gadget/libcomposite.ko"
insmod "/modules/drivers/usb/gadget/function/usb_f_mass_storage.ko
insmod /modules/g_mass_storage.ko file="/data/storage/onboard" removable=y stall=0

(edited)

Szybet

so if i want add audio support to inkbox i need to fork the entire project right?

yeah, basically?

4:11 PM

But you have the release private key, so you can do pretty much all you want (edited)

you wouldn't allow a switch in the settings that adds / dissables some 3 buttons in the ereader while reading, and a app to control playlists? (edited)

Why not?

becouse its bloat for most users

If I can disable/enable them based on the device, no problem (edited)

well, okay

4:12 PM

are the apps list a json list or smth, or is it hard coded?

But make them in Qt

Szybet

are the apps list a json list or smth, or is it hard coded?

No, sorry ;p

4:12 PM

Hardcoded

tux-linux

No, sorry ;p

this needs to be changed

But it's not very difficult to add one

tux-linux

But make them in Qt

yes i know, but the app maybe not

Add a button and a label

tux-linux

But it's not very difficult to add one

it would be still easier to add a app without reinstalling the gui budle

Szybet

yes i know, but the app maybe not

If you do it in X11, it isn't hardcoded

4:13 PM

You can make your extension

4:13 PM

And it will show up automatically in KoBox apps

i think making it bare metal fbink would be better for performance?

yes

so i will do a bare metal

But Qt would be more flexible than FBInk honestly

4:14 PM

It already has all the widgets you need

4:15 PM

And don't even think of doing your app in Python, if that was what you were planning to

tux-linux

And it will show up automatically in KoBox apps

so there are automatically showed apps for x11 or how?

tux-linux

And don't even think of doing your app in Python, if that was what you were planning to

God forbid no python

Python in embedded device is a PITA, because first it isn't a compiled language ok yeah you know the rest

rust

good choice (edited)

4:15 PM

Though I won't help you in cross-compiling rust ;p

the plato project already made it

This handles user extensions https://github.com/Kobo-InkBox/kernel/blob/0fc7867d587b651f901756063d4c14417b344bb0/initrd/common/startx#L215-L298

tux-linux

insmod "/modules/fs/configfs/configfs.ko"
insmod "/modules/drivers/usb/gadget/libcomposite.ko"
insmod "/modules/drivers/usb/gadget/function/usb_f_mass_storage.ko
insmod /modules/g_mass_storage.ko file="/data/storage/onboard" removable=y stall=0

(edited)

Does that work?

kobo:~# insmod "/modules/fs/configfs/configfs.ko"
insmod: ERROR: could not insert module /modules/fs/configfs/configfs.ko: File exists
kobo:~# [ 3995.928562] ==== DISABLE ibat ====

kobo:~# insmod "/modules/drivers/usb/gadget/libcomposite.ko"
insmod: ERROR: could not insert module /modules/drivers/usb/gadget/libcomposite.ko: File exists
kobo:~# insmod "/modules/drivers/usb/gadget/function/usb_f_mass_storage.ko
> ^C
kobo:~# insmod "/modules/drivers/usb/gadget/function/usb_f_mass_storage.ko"
kobo:~# insmod "/modules/drivers/usb/gadget/legacy/g_mass_storage.ko"
[ 4013.100897] no file given for LUN0
[ 4013.104348] g_mass_storage ci_hdrc.0: failed to start g_mass_storage: -22
insmod: ERROR: could not insert module /modules/drivers/usb/gadget/legacy/g_mass_storage.ko: Invalid parameters
kobo:~# insmod "/modules/drivers/usb/gadget/legacy/g_mass_storage.ko"
[ 4021.057430] no file given for LUN0
[ 4021.062868] g_mass_storage ci_hdrc.0: failed to start g_mass_storage: -22
insmod: ERROR: could not insert module /modules/drivers/usb/gadget/legacy/g_mass_storage.ko: Invalid parameters
kobo:~# [ 4025.062738] ==== DISABLE ibat ====
 * WARNING: you are stopping a boot service
 * Unmounting onboard storage filesystem ...
 [ ok ]
 * WARNING: you are stopping a boot service
 * Stopping USBNet ...
rmmod: ERROR: Module /modules/g_ether is not currently loaded
 [ !! ]
 * ERROR: usbnet failed to stop
[ 4054.385494] ==== DISABLE ibat ====

4:19 PM

no*** usb storage on the pc (edited)

ok nice

so it does not work

Well there's progress

yes

rmmod all the modules you have loaded now

can i just reboot?

no

4:22 PM

because it will load them again with usbnet

tux-linux

insmod "/modules/fs/configfs/configfs.ko"
insmod "/modules/drivers/usb/gadget/libcomposite.ko"
insmod "/modules/drivers/usb/gadget/function/usb_f_mass_storage.ko
insmod /modules/g_mass_storage.ko file="/data/storage/onboard" removable=y stall=0

(edited)

I edited the commands there so that it can work

kobo:~# lsmod
Module                  Size  Used by
usb_f_mass_storage     38079  0
libcomposite           44430  1 usb_f_mass_storage
configfs               23581  3 libcomposite,usb_f_mass_storage
8189fs               1294344  0
sdio_wifi_pwr            593  0
kobo:~#

so i: rmod usb_f_mass_storage rmod libcomposite rmod configfs right?

All of them

oh

But yeah, the ones you suggested are the most relevant

kobo:~# insmod "/modules/fs/configfs/configfs.ko"
kobo:~# insmod "/modules/drivers/usb/gadget/libcomposite.ko"
kobo:~# insmod "/modules/drivers/usb/gadget/function/usb_f_mass_storage.ko"
kobo:~# insmod /modules/g_mass_storage.ko file="/data/storage/onboard" removable
=y stall=0
insmod: ERROR: could not load module /modules/g_mass_storage.ko: No such file or directory
kobo:~# insmod /modules/g_mass_storage.ko file="/data/storage/onboard" removable
=y stall=0
insmod: ERROR: could not load module /modules/g_mass_storage.ko: No such file or directory
kobo:~#

hmm

4:26 PM

also, the 3 line miss a "

1

oh

4:27 PM

change /modules/g_mass_storage.ko to /modules/drivers/usb/gadget/legacy/g_mass_storage.ko

4:28 PM

now it works

Nice!

could you describe what you needed to do? (edited)

Just load the necessary modules When on newer kernels, there are dependencies to make the function itself more portable instead of just throwing everything into one module like on older devices

it adds more complecity

4:32 PM

but okay

1

Normally it's handled by modprobe

oh, okay

But when you have it scripted up it isn't a problem anymore

that makes sense

4:34 PM

maybe the wifi choosing menu a little bigger? its smolllll

Szybet

maybe the wifi choosing menu a little bigger? its smolllll

I know right

when i choose wifi, or a file i don't need to see anything else

There isn't any way I found to do this

tux-linux

I know right

good

Maybe it will be slightly bigger in the new UI bundle I made

its qt right?

I'll upload it now

4:35 PM

yeah

4:35 PM

sizePolicy things...

not good

4:35 PM

but still

ok uploaded

4:37 PM

http://23.163.0.39/bundles/private/update-n306.isa

4:37 PM

Download this, copy it in /data/firmware/update.isa sync and reboot

are you sure? on which partition is it?

p4

4:41 PM

/data/firmware is a symlink to [p4]/update

what should i test

4:43 PM

now

4:46 PM

usb storage does not work

Did you update the bundle?

i replaced the file

4:49 PM

so yes

And restarted the device?

i inserted it using u-boot and the booted

Ok

4:50 PM

So is the dpi set to 195?

tux-linux

cat /data/config/09-dpi/config?

could i get privileges to pin messages, that i wouldn't need to scroll

4:51 PM

yes it is 195

I tried to allow you but didn't find how

4:53 PM

Ok found it

4:53 PM

Try now (edited)

Szybet pinned a message to this channel. 3/6/2022 4:53 PM

yes

Now open settings

4:54 PM

Close settings

4:54 PM

And check dpi again

the same

Ok good

5:00 PM

Pull kernel repo (edited)

5:01 PM

I made a symlink for g_mass_storage in modules

5:01 PM

Recompile and flash it

what is the max kernel size?

~10.5M (edited)

adding usb audio, and alsa support could take it

5:07 PM

not good

An alternative would maybe be to put the kernel in a partition and extload it from U-Boot

5:08 PM

Or put the modules... somewhere else

tux-linux

Or put the modules... somewhere else

this this this this

(well, alsa modules I mean)

well this too

5:08 PM

usb storage does not work

Because the tricky part is that the modules differ for each device

Szybet

usb storage does not work

not normal (edited)

5:09 PM

find /modules

kobo:~# find /modules
/modules
/modules/drivers
/modules/drivers/mmc
/modules/drivers/mmc/card
/modules/drivers/mmc/card/sdio_wifi_pwr.ko
/modules/drivers/usb
/modules/drivers/usb/gadget
/modules/drivers/usb/gadget/function
/modules/drivers/usb/gadget/function/u_ether.ko
/modules/drivers/usb/gadget/function/usb_f_acm.ko
/modules/drivers/usb/gadget/function/usb_f_ecm.ko
/modules/drivers/usb/gadget/function/usb_f_ecm_subset.ko
/modules/drivers/usb/gadget/function/usb_f_eem.ko
/modules/drivers/usb/gadget/function/usb_f_fs.ko
/modules/drivers/usb/gadget/function/usb_f_mass_storage.ko
/modules/drivers/usb/gadget/function/usb_f_obex.ko
/modules/drivers/usb/gadget/function/usb_f_rndis.ko
/modules/drivers/usb/gadget/legacy
/modules/drivers/usb/gadget/legacy/g_ether.ko
/modules/drivers/usb/gadget/legacy/g_mass_storage.ko
/modules/drivers/usb/gadget/libcomposite.ko
/modules/fs
/modules/fs/configfs
/modules/fs/configfs/configfs.ko
/modules/g_mass_storage.ko
/modules/wifi
/modules/wifi/8189fs.ko

Oh I know

5:09 PM

I only insmod usb_f_mass_storage in inkbox

5:09 PM

which is not enough

5:10 PM

hang on; I'll change that

5:10 PM

lsmod?

kobo:~# 
kobo:~# lsmod
Module                  Size  Used by
kobo:~# lsmod
Module                  Size  Used by
kobo:~# lsmod
Module                  Size  Used by
kobo:~# lsmod
Module                  Size  Used by
kobo:~# lsmod
Module                  Size  Used by
kobo:~# lsmod
Module                  Size  Used by
kobo:~# lsmod
Module                  Size  Used by
kobo:~# lsmod
Module                  Size  Used by
kobo:~# lsmod
Module                  Size  Used by
kobo:~# lsmod
Module                  Size  Used by
kobo:~#

5:10 PM

HMMMMMMMMM

What

exactly

rc-service inkbox_gui restart

5:11 PM

And your will see insmod output on stdout

Strange error with gled Anyways Try hitting connect button

5:12 PM

In GUI

5:13 PM

i will reboot

5:13 PM

and check one more time

1

5:17 PM

the same result

nothing on lsmod?

nothing

Please generate system report

5:20 PM

Settings -> Page 2 -> Generate system report

can i do this throught cmd?

No

5:20 PM

Well

5:20 PM

maybe

5:20 PM

/usr/local/bin/generate_system_report.sh

5:21 PM

yeah you can actually

5:21 PM

share the pdf file

?

it will produce a pdf in onboard storage

5:23 PM

you will get errors but those are harmless

sysreport-2012-01-01_16.26.46.pdf

248.42 KB

your RTC needs to sync ;p

tux-linux

your RTC needs to sync ;p

all ereaders have problems with handling rtc

5:25 PM

it sucks

yeah

5:26 PM

I see nothing wrong in the report

5:27 PM

What if you try insmod-ing the modules yourself?

Szybet

kobo:~# insmod "/modules/fs/configfs/configfs.ko"
kobo:~# insmod "/modules/drivers/usb/gadget/libcomposite.ko"
kobo:~# insmod "/modules/drivers/usb/gadget/function/usb_f_mass_storage.ko"
kobo:~# insmod /modules/g_mass_storage.ko file="/data/storage/onboard" removable
=y stall=0
insmod: ERROR: could not load module /modules/g_mass_storage.ko: No such file or directory
kobo:~# insmod /modules/g_mass_storage.ko file="/data/storage/onboard" removable
=y stall=0
insmod: ERROR: could not load module /modules/g_mass_storage.ko: No such file or directory
kobo:~#

hmm

those?

yeah

usb storage works

ok

dmesg?

message.txt

57.14 KB

5:32 PM

here is everything

Ok, seems good

5:35 PM

inkbox

5.91 MB

5:35 PM

On host:

cd ~/Downloads
busybox httpd -p 8000 -f

On Kobo:

rm /tmp/inkbox; wget your.computer.ip.there:8000/inkbox -O /tmp/inkbox; chmod +x /tmp/inkbox; umount -l -f /kobo/mnt/onboard/.adds/inkbox/inkbox-bin; mount --bind /tmp/inkbox /kobo/mnt/onboard/.adds/inkbox/inkbox-bin; killall inkbox-bin inkbox inkbox.sh; env QT_QPA_PLATFORM=kobo chroot /kobo /mnt/onboard/.adds/inkbox/inkbox.sh

5:36 PM

But first, rmmod all modules related to usb mass storage

what is this binary?

This basically bind-mounts a temporary binary for testing

5:37 PM

InkBox binary

5:37 PM

I added this code

tux-linux

On host:

cd ~/Downloads
busybox httpd -p 8000 -f

On Kobo:

rm /tmp/inkbox; wget your.computer.ip.there:8000/inkbox -O /tmp/inkbox; chmod +x /tmp/inkbox; umount -l -f /kobo/mnt/onboard/.adds/inkbox/inkbox-bin; mount --bind /tmp/inkbox /kobo/mnt/onboard/.adds/inkbox/inkbox-bin; killall inkbox-bin inkbox inkbox.sh; env QT_QPA_PLATFORM=kobo chroot /kobo /mnt/onboard/.adds/inkbox/inkbox.sh

it does not have wifi, so net over usb is needed?

yeah

5:38 PM

can't you use udhcpc?

5:39 PM

try to connect, let dhcpcd fail and then connect with udhcpc

5:39 PM

Anyway I changed the scripts in the rootfs I'll push that once we fix mass storage

Szybet pinned a message to this channel. 3/6/2022 5:39 PM

5:41 PM

not funny

5:41 PM

maybe i will just push the binary using u-boot?

did you /usr/local/bin/wifi/connect_to_wifi.sh before?

5:41 PM

question: is your power light still flashing?

tux-linux

question: is your power light still flashing?

no

no really? how-

5:43 PM

ls -shal /sys/devices/platform/leds/leds/GLED

okay it is

5:43 PM

im just blind

ah

5:43 PM

okay

5:43 PM

echo 1 > /sys/devices/platform/leds/leds/GLED/brightness; echo 0 > /sys/devices/platform/leds/leds/GLED/brightness turns it off right?

Szybet

Click to see attachment 🖼️

That isn't possible

tux-linux

echo 1 > /sys/devices/platform/leds/leds/GLED/brightness; echo 0 > /sys/devices/platform/leds/leds/GLED/brightness turns it off right?

yes

find /usr/local/bin

Szybet

yes

Ok, I forgot to put brightness in the path in inkbox.sh

kobo:~# find /usr/local/bin/                                                                   
/usr/local/bin/                                                                                
/usr/local/bin/htmldoc                                                                         
/usr/local/bin/ota_update.sh                                                                   
/usr/local/bin/unionfs                                                                         
/usr/local/bin/wifi                                                                            
/usr/local/bin/wifi/toggle.sh                                                                  
/usr/local/bin/wifi/connect_to_network.sh                                                      
/usr/local/bin/wifi/list_networks.sh                                                           
/usr/local/bin/unionfsctl                                                                      
/usr/local/bin/generate_system_report.sh                                                       
/usr/local/bin/gutenberg                                                                       
/usr/local/bin/gutenberg/fetch_catalog.sh                                                      
/usr/local/bin/gutenberg/get_book.sh                                                           
/usr/local/bin/gutenberg/get_latest_books.sh                                                   
/usr/local/bin/gutenberg/search_catalog.sh                                                     
/usr/local/bin/zsh-5.8.1                                                                       
/usr/local/bin/htop                                                                            
/usr/local/bin/zsh                                                                             
/usr/local/bin/appvnc_start.sh

5:44 PM

connect to network, not wifi

/usr/local/bin/wifi/connect_to_network.sh

+ ssid and password

but why couldn't it launch before?

tux-linux

but why couldn't it launch before?

wrong binary name you gived? (edited)

5:46 PM

kobo:~# /usr/local/bin/wifi/connect_to_network.sh hotspotttssss 12345678                       
ifconfig: ioctl 0x8913 failed: No such device                                                  
Successfully initialized wpa_supplicant                                                        
rfkill: Cannot get wiphy information                                                           
ioctl[SIOCSIWAP]: Operation not permitted                                                      
no such user dhcpcd                                                                            
dhcpcd-9.4.0 starting                                                                          
eth0: waiting for carrier                                                                      
timed out                                                                                      
dhcpcd exited                                                                                  
dhcpcd_fork_cb: truncated read 0 (expected 4)                                                  
DHCP request failed.                                                                           
kobo:~# ip a
1: lo: <LOOPBACK,UP,LOWER_UP> mtu 65536 qdisc noqueue state UNKNOWN group default              
    link/loopback 00:00:00:00:00:00 brd 00:00:00:00:00:00                                      
    inet 127.0.0.1/8 scope host lo                                                             
       valid_lft forever preferred_lft forever                                                 
kobo:~# udhcpc -i eth0
udhcpc: ioctl 0x8933 failed: No such device                                                    
kobo:~#

5:47 PM

i need to insmod the wifi module too

5:47 PM

what was his name

/modules/wifi/8189fs.ko

5:48 PM

but

5:48 PM

it shouldn't do that

5:48 PM

cat /usr/local/bin/wifi/connect_to_network.sh

still no eth0

5:48 PM

message.txt

7.32 KB

dmesg and lsmod please

5:49 PM

this is scary...

message.txt

259.73 KB

what if you do insmod /modules/drivers/mmc/card/sdio_wifi_pwr.ko

5:51 PM

oh ok

\kobo:~# insmod /modules/drivers/mmc/card/sdio_wifi_pwr.ko                                      
kobo:~# lsmod                                                                                  
Module                  Size  Used by                                                          
sdio_wifi_pwr            593  0                                                                
8189fs               1294344  0                                                                
usb_f_mass_storage     38079  0                                                                
libcomposite           44430  1 usb_f_mass_storage                                             
configfs               23581  3 libcomposite,usb_f_mass_storage                                
kobo:~# ip a                                                                                   
1: lo: <LOOPBACK,UP,LOWER_UP> mtu 65536 qdisc noqueue state UNKNOWN group default              
    link/loopback 00:00:00:00:00:00 brd 00:00:00:00:00:00                                      
    inet 127.0.0.1/8 scope host lo                                                             
       valid_lft forever preferred_lft forever                                                 
5: eth0: <BROADCAST,MULTICAST> mtu 1500 qdisc noop state DOWN group default qlen 1000          
    link/ether 58:b0:d4:9d:a6:8e brd ff:ff:ff:ff:ff:ff                                         
kobo:~# 
\

5:51 PM

bingo

Actually it auto-removes the modules upon failure to connect in the script

wtf

message.txt

5.07 KB

5:53 PM

it dissapered

5:53 PM

kobo:~# lsmod                                                                                  
Module                  Size  Used by                                                          
usb_f_mass_storage     38079  0                                                                
libcomposite           44430  1 usb_f_mass_storage                                             
configfs               23581  3 libcomposite,usb_f_mass_storage                                
kobo:~#

5:53 PM

dmesg:

message.txt

95.11 KB

5:54 PM

wifi 3v3 off

5:54 PM

weird

Ok, I'll push a commit to the rootfs repo so you can get out of this mess with proper scripts

5:56 PM

Done

pull rootfs repo, make a sqsh and sign, then put in on kobo?

yes

Szybet pinned a message to this channel. 3/6/2022 5:59 PM

Szybet pinned a message to this channel. 3/6/2022 6:01 PM

done, what now

Try connecting to Wi-Fi with the script again (edited)

so /usr/local/bin/wifi/connect_to_network.sh hotspotttssss 12345678 and then udhcpc -i eth0 ?

yep

kobo:~# ip a                                                                                   
1: lo: <LOOPBACK,UP,LOWER_UP> mtu 65536 qdisc noqueue state UNKNOWN group default              
    link/loopback 00:00:00:00:00:00 brd 00:00:00:00:00:00                                      
    inet 127.0.0.1/8 scope host lo                                                             
       valid_lft forever preferred_lft forever                                                 
2: eth0: <BROADCAST,MULTICAST,UP,LOWER_UP> mtu 1500 qdisc mq state UP group default qlen 1000  
    link/ether 58:b0:d4:9d:a6:8e brd ff:ff:ff:ff:ff:ff                                         
    inet 10.42.0.28/24 brd 10.42.0.255 scope global eth0                                       
       valid_lft forever preferred_lft forever

6:09 PM

yep, now what

very good, it works

6:10 PM

try with the inkbox binary above

tux-linux

On host:

cd ~/Downloads
busybox httpd -p 8000 -f

On Kobo:

rm /tmp/inkbox; wget your.computer.ip.there:8000/inkbox -O /tmp/inkbox; chmod +x /tmp/inkbox; umount -l -f /kobo/mnt/onboard/.adds/inkbox/inkbox-bin; mount --bind /tmp/inkbox /kobo/mnt/onboard/.adds/inkbox/inkbox-bin; killall inkbox-bin inkbox inkbox.sh; env QT_QPA_PLATFORM=kobo chroot /kobo /mnt/onboard/.adds/inkbox/inkbox.sh

this

Szybet

kobo:~# ip a                                                                                   
1: lo: <LOOPBACK,UP,LOWER_UP> mtu 65536 qdisc noqueue state UNKNOWN group default              
    link/loopback 00:00:00:00:00:00 brd 00:00:00:00:00:00                                      
    inet 127.0.0.1/8 scope host lo                                                             
       valid_lft forever preferred_lft forever                                                 
2: eth0: <BROADCAST,MULTICAST,UP,LOWER_UP> mtu 1500 qdisc mq state UP group default qlen 1000  
    link/ether 58:b0:d4:9d:a6:8e brd ff:ff:ff:ff:ff:ff                                         
    inet 10.42.0.28/24 brd 10.42.0.255 scope global eth0                                       
       valid_lft forever preferred_lft forever

Are you using a Wi-Fi hotspot with Gnome?

6:11 PM

Mine has the same subnet than yours ;p

its kde

6:11 PM

they use the same thing for this propably

ah

6:11 PM

Well, networkmanager, so makes sense, yeah

its actually something else

6:12 PM

i looked at it once

oh?

its called dnsmasq

6:14 PM

in htop you can see the command arguments for it

6:14 PM

% ps aux | grep dnsmasq                                                      
nobody    307584  0.0  0.0  21688  6808 ?        S    15:23   0:00 /usr/bin/dnsmasq --conf-file=/dev/null --no-hosts --keep-in-foreground --bind-interfaces --except-interface=lo --clear-on-reload --strict-order --listen-address=10.42.0.1 --dhcp-range=10.42.0.10,10.42.0.254,60m --dhcp-lease-max=50 --dhcp-leasefile=/var/lib/NetworkManager/dnsmasq-wlp3s0.leases --pid-file=/var/run/nm-dnsmasq-wlp3s0.pid --conf-dir=/etc/NetworkManager/dnsmasq-shared.d
szybet    482350  0.0  0.0   9548  2560 pts/7    S+   18:14   0:00 grep --color=auto --exclude-dir=.bzr --exclude-dir=CVS --exclude-dir=.git --exclude-dir=.hg --exclude-dir=.svn --exclude-dir=.idea --exclude-dir=.tox dnsmasq
18:14 szybet@ArchSzybet /home/szybet

6:14 PM

i have the inkbox test binary

6:14 PM

launched

6:15 PM

what now, usb storage?

Szybet

% ps aux | grep dnsmasq                                                      
nobody    307584  0.0  0.0  21688  6808 ?        S    15:23   0:00 /usr/bin/dnsmasq --conf-file=/dev/null --no-hosts --keep-in-foreground --bind-interfaces --except-interface=lo --clear-on-reload --strict-order --listen-address=10.42.0.1 --dhcp-range=10.42.0.10,10.42.0.254,60m --dhcp-lease-max=50 --dhcp-leasefile=/var/lib/NetworkManager/dnsmasq-wlp3s0.leases --pid-file=/var/run/nm-dnsmasq-wlp3s0.pid --conf-dir=/etc/NetworkManager/dnsmasq-shared.d
szybet    482350  0.0  0.0   9548  2560 pts/7    S+   18:14   0:00 grep --color=auto --exclude-dir=.bzr --exclude-dir=CVS --exclude-dir=.git --exclude-dir=.hg --exclude-dir=.svn --exclude-dir=.idea --exclude-dir=.tox dnsmasq
18:14 szybet@ArchSzybet /home/szybet

oh cool

6:15 PM

yes

6:15 PM

try hitting "connect"

6:15 PM

then dmesg and lsmod

does not work

message.txt

93.89 KB

Well, it actually does something, great

something

[  651.464636] g_mass_storage: Unknown symbol fsg_common_set_ops (err 0)
[  651.464680] g_mass_storage: Unknown symbol fsg_common_set_sysfs (err 0)
[  651.464704] g_mass_storage: Unknown symbol usb_get_function_instance (err 0)
[  651.464716] g_mass_storage: Unknown symbol fsg_common_run_thread (err 0)
[  651.464734] g_mass_storage: Unknown symbol usb_string_ids_tab (err 0)
[  651.464751] g_mass_storage: Unknown symbol fsg_common_free_luns (err 0)
[  651.464763] g_mass_storage: Unknown symbol usb_get_function (err 0)
[  651.464776] g_mass_storage: Unknown symbol usb_composite_unregister (err 0)
[  651.464788] g_mass_storage: Unknown symbol usb_put_function (err 0)
[  651.464800] g_mass_storage: Unknown symbol fsg_common_create_luns (err 0)
[  651.464813] g_mass_storage: Unknown symbol usb_add_function (err 0)
[  651.464826] g_mass_storage: Unknown symbol usb_composite_probe (err 0)
[  651.464842] g_mass_storage: Unknown symbol usb_composite_overwrite_options (err 0)
[  651.464854] g_mass_storage: Unknown symbol usb_otg_descriptor_alloc (err 0)
[  651.464867] g_mass_storage: Unknown symbol usb_put_function_instance (err 0)
[  651.464878] g_mass_storage: Unknown symbol fsg_common_set_cdev (err 0)
[  651.464892] g_mass_storage: Unknown symbol fsg_common_free_buffers (err 0)
[  651.464905] g_mass_storage: Unknown symbol usb_otg_descriptor_init (err 0)
[  651.464921] g_mass_storage: Unknown symbol fsg_common_set_nluns (err 0)
[  651.464934] g_mass_storage: Unknown symbol fsg_common_remove_luns (err 0)
[  651.464945] g_mass_storage: Unknown symbol usb_add_config (err 0)
[  651.464958] g_mass_storage: Unknown symbol fsg_common_set_inquiry_string (err 0)
[  651.464970] g_mass_storage: Unknown symbol fsg_common_set_num_buffers (err 0)
[  651.464982] g_mass_storage: Unknown symbol fsg_config_from_params (err 0)

something

So you're sure there's nothing in lsmod right?

kobo:~# lsmod
Module                  Size  Used by
8189fs               1294344  0
sdio_wifi_pwr            593  0
kobo:~#

cat /kobo/opt/inkbox_device

6:18 PM

ok

emu

6:18 PM

why emu

6:18 PM

D:

aaahaah

XD

echo n306 > /kobo/opt/inkbox_device

and?

relaunch inkbox and things will probably be better

hm, how. i have the custom binary launched.... just /tmp/inkbox?

6:20 PM

and kill it before launching

just re-launch the big command

6:20 PM

it will do everything for you

6:20 PM

it's the one I use for quick testing

tux-linux pinned a message to this channel. 3/6/2022 6:20 PM

kobo:~# rm /tmp/inkbox; wget 10.42.0.1:8000/inkbox -O /tmp/inkbox; chmod +x /tmp               
/inkbox; umount -l -f /kobo/mnt/onboard/.adds/inkbox/inkbox-bin; mount --bind /t               
^[[A^[[B^[[A^[[B^C
kobo:~# ^C
insmod: can't insert '/external_root/modules/fs/configfs/configfs.ko': File exists
insmod: can't insert '/external_root/modules/drivers/usb/gadget/libcomposite.ko': File exists
insmod: can't insert '/external_root/modules/drivers/usb/function/usb_f_mass_storage.ko': No such file or directory

6:22 PM

and it does not work

ls /kobo/external_root/modules/drivers/usb/function/usb_f_mass_storage.ko

6:22 PM

well, it's progress!

6:22 PM

File exists mean the module is already loaded

tux-linux

File exists mean the module is already loaded

kobo:~# lsmod
Module                  Size  Used by
libcomposite           44430  0
configfs               23581  2 libcomposite
8189fs               1294344  0
sdio_wifi_pwr            593  0
kobo:~#

yep they are

6:23 PM

oh my bad I know the problem

kobo:~# ls /kobo/external_root/modules/drivers/usb/function/usb_f_mass_storage.ko
ls: /kobo/external_root/modules/drivers/usb/function/usb_f_mass_storage.ko: No such file or directory
kobo:~# ls /kobo/external_root/modules/drivers/usb/gadget/
function/        legacy/          libcomposite.ko
kobo:~# ls /kobo/external_root/modules/drivers/usb/gadget/
function/        legacy/          libcomposite.ko
kobo:~# ls /kobo/external_root/modules/drivers/usb/gadget/function/u
u_ether.ko             usb_f_ecm_subset.ko    usb_f_mass_storage.ko
usb_f_acm.ko           usb_f_eem.ko           usb_f_obex.ko
usb_f_ecm.ko           usb_f_fs.ko            usb_f_rndis.ko
kobo:~# ls /kobo/external_root/modules/drivers/usb/gadget/function/usb_f_mass_storage.ko 
/kobo/external_root/modules/drivers/usb/gadget/function/usb_f_mass_storage.ko
kobo:~#

6:23 PM

yes you know

6:23 PM

wrong path

try this one

inkbox

5.91 MB

usb storage works

yay!

still some errors:

cat: can't open '/mnt/onboard/.kobo/Kobo/Kobo eReader.conf': No such file or directory
insmod: can't insert '/external_root/modules/fs/configfs/configfs.ko': File exists                             
insmod: can't insert '/external_root/modules/drivers/usb/gadget/libcomposite.ko': File exists

it's normal, since the configfs and libcomposite modules were already loaded

okay

I'll make you a GUI bundle with this binary

don't forget to update the device from emu to n306

yes

when wifi will work?

https://inkbox.ddns.net/git/tux-linux/rootfs/-/commit/5cba773fdbdf36637195be0488cae34383159670

inkbox_gui: Bind-mount 'device' file to GUI chroot (5cba773f) · Com...

InkBox OS root filesystem

Szybet

when wifi will work?

it doesn't?

without using cmd

6:32 PM

only GUI

try it

6:33 PM

it uses the same script as if you were in cmd

6:35 PM

You can use htop to see what it's doing

6:35 PM

maybe top if you're in a serial console

usb storage doesn't disconnect after i unmount it :(

6:36 PM

bruh sdd 8:48 1 0B 0 disk

You need to unplug the cable

why? nickel doesn't needed

because I haven't found a way to detect if it was ejected or not

hmm, interesting

6:40 PM

it works

6:40 PM

how inkbox manages turning off the wifi?

6:40 PM

how do i get the IP of the device...

6:41 PM

:|

Settings -> page 2 -> system info

6:43 PM

Get to the bottom

WHY

It will be there

to get the ip i need to generate a whole system report

No

6:43 PM

Just system info

6:44 PM

Not the same thing

still... no

It's a dialog

show a window after connecting to wifi (edited)

6:44 PM

with the ip

you can do nmap -sP 10.42.0.0/24 on your host otherwise

tux-linux

you can do nmap -sP 10.42.0.0/24 on your host otherwise

i can even check it with cmd on serial connection

6:45 PM

but thats not the case

Szybet

how inkbox manages turning off the wifi?

and this?

it doesn't

6:45 PM

maybe this could be something to add in settings

tux-linux

it doesn't

so it just disconnects randomly?

no

6:46 PM

it stays connected

it disconnected

huh?

6:47 PM

dmesg?

let me check this one more time

6:47 PM

do i need to enter the password every time?

normally, no

but?

it auto-connects on boot

6:48 PM

it's complicated, but it will only do so when we have the recoveryfs

6:48 PM

i.e. in release

i see here room for improvement, its a basic feature to remember wifi passwords

it does remember them

i need to enter it one more time now

6:49 PM

so how does it remember

cat /data/config/17-wifi_connection_information/*

yes, there are stored password and essid

6:50 PM

but i needed to enter it one more time

The main problem with auto-connecting to a network would be if you changed the password of the ssid or had the same ssid but not the same password

6:51 PM

Once the recoveryfs is in place, the initrd will do that on boot: https://github.com/Kobo-InkBox/kernel/blob/master/initrd/common/setup-wifi (edited)

tux-linux

The main problem with auto-connecting to a network would be if you changed the password of the ssid or had the same ssid but not the same password

a button to reset the password and essid

But then it has to see if the ssid is the same as the one saved, open another dialog i.e. pain ;p

6:53 PM

but it could be doable

6:53 PM

Though it would not have much purpose unless you decided to disable Wi-Fi and re-enable it after

6:54 PM

since it would already be auto-connected on boot

and a button to enable, disable the wifi

that is even more complicated

tux-linux

But then it has to see if the ssid is the same as the one saved, open another dialog i.e. pain ;p

its more pain to enter the password every time

the Wi-Fi connection framework was one of the hardest and most confusing things I have done in Qt

6:56 PM

but you are connected now, it works, right?

yes it works, and it does not dissconnect

6:57 PM

but still pain to enter the password every time

6:57 PM

or maybe a dialog "would you like to enter a new password or use this: "xyz""

Hmm yeah maybe

and a list of saved networks + password to show

6:58 PM

I really will have to learn cpp and qt

Can you send me a video of how InkBox behaves on the Nia? Basic things, like reading a book, searching something, online library?

6:59 PM

So I can ensure everything seems fine on that model?

Szybet

and a list of saved networks + password to show

Wowow, one network at a time

InkBox only saves one for the moment

online library? what?

7:01 PM

you mean gutenberg, libraru button?

Yes

um error

Whoops

This plugin does not support setParent!                                                                        
Gutenberg sync encountered an error

First one is unrelated

7:02 PM

Is there any other output?

nope

ls /data/storage/gutenberg -shal

kobo:~# ls /data/storage/gutenberg -shal total 12K 4 drwxr-xr-x 2 root root 4.0K Jan 1 18:08 . 4 drwxr-xr-x 10 root root 4.0K Jan 1 00:54 .. 0 -rw-r--r-- 1 root root 0 Jan 1 18:08 catalog.csv 4 -rw-r--r-- 1 root root 11 Jan 1 18:08 last_sync kobo:~#

Can you ping Google?

yes

sh -x /usr/local/bin/gutenberg/fetch_catalog.sh

kobo:~# sh -x /usr/local/bin/gutenberg/fetch_catalog.sh                                                        
+ mkdir -p /data/storage/gutenberg                                                                             
+ cd /data/storage/gutenberg                                                                                   
+ date '+%s'                                                                                                   
+ rm -f catalog.csv                                                                                            
+ wget -O catalog.csv http://www.gutenberg.org/cache/epub/feeds/pg_catalog.csv                                 
--2012-01-01 18:11:41--  http://www.gutenberg.org/cache/epub/feeds/pg_catalog.csv                              
Resolving www.gutenberg.org... 152.19.134.47, 2610:28:3090:3000:0:bad:cafe:47                                  
Connecting to www.gutenberg.org|152.19.134.47|:80... connected.                                                
HTTP request sent, awaiting response... 302 Found                                                              
Location: https://www.gutenberg.org/cache/epub/feeds/pg_catalog.csv [following]                                
--2012-01-01 18:11:42--  https://www.gutenberg.org/cache/epub/feeds/pg_catalog.csv                             
Connecting to www.gutenberg.org|152.19.134.47|:443... connected.                                               
ERROR: cannot verify www.gutenberg.org's certificate, issued by 'CN=Network Solutions OV Server CA 2,O=Network:
  Issued certificate not yet valid.                                                                            
To connect to www.gutenberg.org insecurely, use `--no-check-certificate'.                                      
+ exit 1                                                                                                       
kobo:~#

7:05 PM

o h n o c e r t i f i c a t e

i feel that pain

Sync your rtc

with?

7:12 PM

in alpine there are like 5 diffrent

7:12 PM

the stock one was not working

7:12 PM

so idk which you used (edited)

date -s "2022-03-06 19:17

7:17 PM

hwclock --systohc

7:17 PM

and you're done

7:18 PM

it's not a real sync, but the simplest method I have (edited)

maybe a NTP would be better

Or this

7:18 PM

sudo date -s "$(wget --method=HEAD -qSO- --max-redirect=0 google.com 2>&1 | grep Date: | cut -d' ' -f2-7)"

7:18 PM

From that question https://unix.stackexchange.com/questions/79112/how-do-i-set-time-and-date-from-the-internet

How do I set time and date from the Internet?

How do I retrieve the date from the Internet and set my computer's clock, from the command line?

oh thats cool

don't forget to hwclock --systohc after that

it takes long

7:21 PM

but no error soo fat

7:21 PM

kobo:~# Gutenberg sync successfully completed

nice

7:23 PM

Does something show up when you click on "Library"?

books

nice

boooooks

booooooooooks

7:24 PM

have a screenshot?

i will make the video

1

7:24 PM

what do you mean by search

Well, the search button

7:25 PM

Another thing to try after that would be X11

7:25 PM

It should be ok but just to be sure

lightmaps are flickering

7:25 PM

not usable at all

well it's usually not the main reason why someone installs inkbox (edited)

but i clicked it accidentally

7:26 PM

ahh

7:26 PM

you can rc-service inkbox_gui restart

7:27 PM

oh; you will need to run again the command with wget that fetches inkbox binary

you mean search a word?

yeah, or a book in online library

7:28 PM

or in your onboard storage

does this search a book local?

to see how the keyboard behaves

Szybet

does this search a book local?

both

yes it does

7:28 PM

okay

7:33 PM

why for every book it shows end page and no previous page

?

wait, i will upload the video

7:33 PM

weird

are you able to read though?

blank page

if you click next

"end of document"

7:35 PM

the gui propably crashed, i cant click anything

7:35 PM

i will restart

problems

7:37 PM

i'll make you an ui bundle

hm

7:37 PM

i rebooted

7:37 PM

the inkbox file was in /tmp

7:38 PM

here is a old inkbox binary

7:38 PM

hmmm.... thats why?

idk

now reading works

currently making a new one

7:39 PM

nice

make a ui bundle, i will test it and then send a video

1

7:39 PM

everything works great exept the file chooser, it takes ages sometimes to load

7:40 PM

changing brightness in the book does not work, i cant do antyhing

Szybet

changing brightness in the book does not work, i cant do antyhing

hmm hmm

i can click the + - but its really hard, and the slider does not work

7:42 PM

now i see

ok, so it works actually

7:42 PM

it's not a slider, it's a progress bar sorry

1 opened a pdf, works 2 opened a epub, blank page cant go back, forward and options doesn't work

7:43 PM

no info on serial

7:43 PM

:|

yeah I had this bug at some point too

and?

don't know what's causing it

7:43 PM

probably some tmpfs mount/umount quirk

only with epubs?

no, when opening a pdf then an epub

7:44 PM

uploaded the bundle

7:44 PM

http://23.163.0.39/bundles/private/update-n306.isa

its not usable that way

reading books doesn't work

7:57 PM

any

7:57 PM

rebooted, opened epub, and nothing is there

wow

7:57 PM

huh

7:58 PM

when you open an epub, do you see mutool in the process list with top or ps?

wait

8:00 PM

it boots really slow btw

what do you mean?

1 minute?

that's not slow (edited)

8:01 PM

Kindles boot in 1m40s

nickel in 30?

yeah, well, different infrastructure, less secure and more crash-prone

message.txt

1.98 KB

8:01 PM

nothing

Does the book actually exist?

yes

8:03 PM

none of my books work

hang on; let me try this latest commit on the aura 2

8:03 PM

if it doesn't work we have a serious problem

yes

ok, will boot it now

8:11 PM

well, mine works alright

8:12 PM

What's the "Git" part of the system info in Settings?

8:13 PM

Mine is 1.8-86-g56db7ea

wtf 1 reboot 2 opened a pdf 3 times 3 opened another pdf, blank page

Do you have the "Global reading settings" option disabled?

Szybet pinned a message to this channel. 3/6/2022 8:16 PM

There's also a tiny shell script around this command, so you can just as well do restart inkbox_gui

8:17 PM

(or any other service)

tux-linux

Do you have the "Global reading settings" option disabled?

no?

8:17 PM

disabled now

try again; I'm curious

8:18 PM

this option, when disabled, permits you to save the last page you read in the book and go back to it automatically

some pdfs work, some not

8:20 PM

only one pdf is working

8:20 PM

only one book is working

8:20 PM

thats the one i used in the past to test things

try one from gutenberg

8:21 PM

they have a very high success rate

8:21 PM

or you can get one from online library

8:21 PM

btw, you can also display images along pdf and epubs

8:21 PM

pngs, jpgs, etc.

i don't have any

heh, discovered that if you open an image, then an epub, then the image again, it segfaults

8:23 PM

awesome

8:24 PM

Some bugs may have appeared like that but since my userbase is currently very tiny I don't know much about them

8:24 PM

I should really implement something like sickel

from gutenberg it works

Which restarts Nickel when it segfaults/closes unexpectedly

Szybet

from gutenberg it works

ok nice

8:26 PM

I tried with other books, even one with 1500+ pages and it worked

8:26 PM

(not from gutenberg)

8:26 PM

Maybe it's specific to your epub

all of my books?

8:27 PM

only one is working, that is some images pasted into pdf using libreoffice

8:27 PM

._ .

can you send one so I can try on my side?

WS2811.pdf

315.73 KB

8:28 PM

This for example

8:28 PM

its a datasheet of a diode...

Let me see

8:30 PM

works here

8:30 PM

can you try to open the system report pdf you generated before?

no, i cant

8:31 PM

xD

btw, thanks for contributing so much

Szybet

no, i cant

? it doesn't work?

tux-linux

btw, thanks for contributing so much

thanks for making this project, it will make it easier to achieve my goal too

1

tux-linux

? it doesn't work?

it doesn't

8:33 PM

thats a interesting error when Option button doesnt work: Segmentation fault (edited)

ah ah (edited)

8:34 PM

what have you done since the reboot?

tryied to open some books you know

hmm

8:34 PM

if you restart inkbox_gui and try to open the system report pdf, does it work?

btw,

kobo:~# ps aux | grep mutool                                                                                   
 3541 root      0:00 grep mutool                                                                               
kobo:~#

its while in the opened pdf that works

no

8:35 PM

it just shows the grep process that greps mutool

8:35 PM

so there is nothing

8:35 PM

strange

yes i know

8:36 PM

but now the system report works?????

magic

let my try the diode

8:37 PM

nope

what-

but after it the systemreport works

but it works here

8:37 PM

nonsense

._.

I opened an EPUB->System Report->Diode->System Report and it works

let me try

like, it should already have failed three times here lol

Szybet

let me try

restart gui before

after epub the systemreport doesn't open

8:39 PM

but i will restart

8:41 PM

it doesn't work after the reset too

even more nonsense

gutenberg doesn't work

8:42 PM

will reset and try the diode

1

8:43 PM

kobo:~# umount: can't unmount /inkbox: Invalid argument                                                        
iniparser: cannot open /mnt/onboard/.kobo/nightmode.ini                                                        
cat: can't open '/mnt/onboard/.kobo/Kobo/Kobo eReader.conf': No such file or directory                         
QCoreApplication::postEvent: Unexpected null receiver                                                          
sha256sum: /kobo/mnt/onboard/onboard/WS2811.pdf: No such file or directory                                     
mount: mounting /opt/config/07-words_number on /kobo/mnt/onboard/.adds/inkbox/.config/07-words_number failed: y
mount: mounting /opt/config/08-recent_books on /kobo/mnt/onboard/.adds/inkbox/.config/08-recent_books failed: y
mount: mounting /opt/config/11-menubar on /kobo/mnt/onboard/.adds/inkbox/.config/11-menubar failed: No such fiy
mount: mounting /opt/config/13-epub_page_size on /kobo/mnt/onboard/.adds/inkbox/.config/13-epub_page_size faily
mount: mounting /opt/config/14-reader_scrollbar on /kobo/mnt/onboard/.adds/inkbox/.config/14-reader_scrollbar y
QIODevice::read (QFile, "/inkbox/book/split/."): device not open                                               
QIODevice::read (QFile, "/inkbox/book/split/.."): device not open

8:43 PM

idk, doesn't work

mount

8:44 PM

Please post output of this command

serial doesn't format properly the output

8:45 PM

wait

Szybet

serial doesn't format properly the output

well, it's serial, what do you want ;p

this could be host controller

idk

kobo:~# umount: can't unmount /inkbox: Invalid argument
iniparser: cannot open /mnt/onboard/.kobo/nightmode.ini
cat: can't open '/mnt/onboard/.kobo/Kobo/Kobo eReader.conf': No such file or directory
QCoreApplication::postEvent: Unexpected null receiver
QCoreApplication::postEvent: Unexpected null receiver
sha256sum: /kobo/mnt/onboard/onboard/WS2811.pdf: No such file or directory
mount: mounting /opt/config/07-words_number on /kobo/mnt/onboard/.adds/inkbox/.config/07-words_number failed: No such file or directory
mount: mounting /opt/config/08-recent_books on /kobo/mnt/onboard/.adds/inkbox/.config/08-recent_books failed: No such file or directory
mount: mounting /opt/config/11-menubar on /kobo/mnt/onboard/.adds/inkbox/.config/11-menubar failed: No such file or directory
mount: mounting /opt/config/13-epub_page_size on /kobo/mnt/onboard/.adds/inkbox/.config/13-epub_page_size failed: No such file or directory
mount: mounting /opt/config/14-reader_scrollbar on /kobo/mnt/onboard/.adds/inkbox/.config/14-reader_scrollbar failed: No such file or directory
QIODevice::read (QFile, "/inkbox/book/split/."): device not open
QIODevice::read (QFile, "/inkbox/book/split/.."): device not open

ok, and mount?

?

Well, just enter mount, then post the output

8:49 PM

no arguments

message.txt

7.08 KB

CSUM=$(sha256sum /data/onboard/WS2811.pdf | awk '{ print $1 }')
ls -shalR "/data/onboard/.${CSUM}"

(edited)

can i have directories?

8:52 PM

/data/onboard/elektronics/WS2811.pdf

8:52 PM

?

yeah

kobo:~# CSUM=$(sha256sum /data/onboard/elektronics/WS2811.pdf  | awk '{ print $1                                         
 }')
ls: /data/onboard/elektronics/.b679e125ca8793e9c2fc5dbe19897c2ca92226409459f2e590b38634a832f59e: No such file or directoy
kobo:~#

interesting

8:54 PM

oh

8:54 PM

no actually, it should be in the root of onboard

n bjfhdnsajk

8:54 PM

fmdakjm

8:54 PM

thats why the one pdf works

you can put books in subdirectories, but the config settings will be on the root of the onboard storage

8:55 PM

or... it should work?

8:55 PM

normally they do

its the only one that works, the pdf is in the main directory

8:55 PM

wait, i will copy the diode

yeah, try that

it works

that's not good news for me

8:57 PM

what if you try the pdf in subdirectory without global reading settings?

they are disabled

8:58 PM

oh, a pdf

8:58 PM

wait

8:58 PM

it works?

8:59 PM

but it doesn't respond now, the screen

8:59 PM

QIODevice::read (QFile, "/inkbox/book/split/."): device not open
QIODevice::read (QFile, "/inkbox/book/split/.."): device not open
QPixmap::scaled: Pixmap is a null pixmap

8:59 PM

i cant do anything

*with global reading settings

9:00 PM

sorry

oh

my fault

its still not good that it crashed

yeah

it works now

ok

9:02 PM

and that also explains why the gutenberg book worked

9:02 PM

now you can make the video ;p

what with epubs?

try if you want

9:03 PM

but inkbox saved the gutenberg book in onboard storage root

they work

9:03 PM

i dont understand anything now

me neither

When the option is disabled, per-book settings are handled as follows: https://github.com/Kobo-InkBox/rootfs/blob/5cba773fdbdf36637195be0488cae34383159670/etc/init.d/ibxd#L86-L103

9:06 PM

(with the bookconfig_mount function at the top of the file too)

some epubs work, some not, i tryied to launch a pdf, and the previous book was launched, then it crashed

9:10 PM

you will se it on the video

9:12 PM

https://anonfiles.com/ddPex6M4x1/VID_20220306_210545_mp4

9:13 PM

dont ask me why the video is rotated

ok, will look at it, thanks!

when will you figure out why are some books not working?

once it... finishes downloading

9:15 PM

i would like to use the ereader finally

Szybet

when will you figure out why are some books not working?

with the global reading settings thing?

you dont need to download it, you can watch it in the browser

no it downloaded it directly

9:16 PM

and it's the same at the end

tux-linux

with the global reading settings thing?

idk, generally

well, your books work now with that option enabled, right?

some yes, some not. the video shows it

ok

ok finally downloaded

9:40 PM

little tip: you can click on the area where you want the slider to go to get there easily

it takes more time

well, that's a depressive video

everything works

9:43 PM

and then nothing

did you experience UI freeze sometimes? I think so, right?

sometimes

9:44 PM

the file chooser, maybe changing pages

though the video shows you stressed the UI to its maximum so it should have crashed already

??? stressed ui?

by freeze, I mean completely freezing, you're not able to do anything, you have to restart inkbox_gui?

Szybet

??? stressed ui?

well, rapidly clicking next page button for example

9:45 PM

Most devices handle that well but on my Libra for example this can lead to an UI freeze

9:45 PM

Since the Libra has the same kernel version that of the Nia, I wanted to check

tux-linux

by freeze, I mean completely freezing, you're not able to do anything, you have to restart inkbox_gui?

through the video, only at the end

tux-linux

well, rapidly clicking next page button for example

i cant click fast? common...

yes, you can, normally

Szybet

through the video, only at the end

I think that UI freeze is due to the fact that it expects to have a PDF displayed but it's actually an epub and it panics, so unrelated

maybe

I was talking about UI freezes by the platform plugin

Szybet

i cant click fast? common...

it works on touch, mini, glo, glo hd, aura 2, but is unstable on the Libra

9:47 PM

so I wanted to see if yours was unstable or not

9:47 PM

it seems stable

yea, it seems

9:48 PM

i will test it more when opening epubs and pdf will be not faulty

1

tux-linux

I think that UI freeze is due to the fact that it expects to have a PDF displayed but it's actually an epub and it panics, so unrelated

Actually, it segfaults there, right?

maybe, i didn't have serial connection then

can you try to reproduce the error? that way we'll be sure

one minute

1

9:54 PM

1 boot up 2 open the epub last time it errored, nothing shows up, last page, no previous page but i can go into options 3 launch a epub that worked before, it works 4 launch the pdf that worked all the time, it freezes exactly like the last book on the video, no information on serial

ok, well first do rc-service inkbox_gui restart to see inkbox stdout

9:58 PM

it is silenced by default otherwise

oh

do you have global reading settings enabled or disabled?

i will check, i hope it saved

1

10:01 PM

enabled

10:01 PM

is it good?

10:06 PM

@tux-linux

Yes

10:44 PM

It is

10:44 PM

Sorry for the delay

10:50 PM

Can you try X11?

10:50 PM

I will hopefully have the courage to debug that book opening issue

10:50 PM

soon™️

1 opened a epub, it doesn't work 2 home 3 opened a pdf, the screen flickered for 2 minutes 4 book menu, nothing works, and: Segmentation fault

ok yeah so it is not related to the QPA, as expected

tux-linux

Can you try X11?

its rebooting

1

Szybet

1 opened a epub, it doesn't work 2 home 3 opened a pdf, the screen flickered for 2 minutes 4 book menu, nothing works, and: Segmentation fault

yay, can reproduce here

but why the first epub didn't worked?

that I don't know

thats not good

was it just after a reboot?

yes

can you share the epub?

sended on pw (edited)

1

10:57 PM

xorg works

10:57 PM

and its maybe faster?

what do you mean?

idk, maybe just the menu

10:58 PM

im trying to launch geany

ok

10:58 PM

can take ~1m

10:59 PM

on newer devices it's usually faster

10:59 PM

kterm can launch in 45s on the libra

black screen

10:59 PM

and

10:59 PM

Server terminated successfully (0). Closing log file

yes

11:00 PM

normal

11:00 PM

wait a bit

11:00 PM

you can see in ps if geany is running

serial cmd is lagging

11:01 PM

kobo:~# free                                                                                   
              total        used        free      shared  buff/cache   available                
Mem:         250128       90136        2712        1696      157280      137400                
Swap:             0           0           0

137M is quite a lot

hm, it will run out of memory

no?

but free

11:01 PM

but ok

No, check the available part

11:01 PM

the others are not relevant

it works, its horrible and no thanks

it's that horrible?

for me, yes

11:02 PM

for you, no

maybe you mean the size of the keyboard

11:02 PM

that is horrible

it flickers too much for me

well, what do you expect, it's an eink screen

a native fbink editor would work better

try kterm

i will write one in the future i hope (edited)

maybe it will be better

Szybet

a native fbink editor would work better

check out inkvt

11:04 PM

echo true > /boot/flags/DISPLAY_DEBUG

11:04 PM

that will throw serial console output on the screen instead of the serial port

11:04 PM

with inkvt

11:05 PM

you can quit geany with file->quit

11:05 PM

gui will launch automatically after that

well, tommorow i will check it

1

ePUB/PDF opening issue was fixed with those two commits: https://github.com/Kobo-InkBox/inkbox/commit/8501b42124a8c44c79dfa455d3dfac577395bb99 https://github.com/Kobo-InkBox/gui-bundle/commit/9fc6551f054dc97be82b9cc316fad32b7e50fb7c

Reader framework: fix file opening bug · Kobo-InkBox/inkbox@8501b42

When an user first opened an ePUB, then a PDF, it didn't work. This was due to an issue with the two tmpfs at '/kobo/inkbox' and '/kobo/run'. This commit fixes that issue.

Add remount_tmpfs_launch.sh script; improve inkbox.sh · Kobo-InkBox...

You can prepare a GUI bundle

6:41 AM

And i think in the future i will make a docker image with Qt and everything to compile inkbox

Nice

6:46 AM

12:40AM here so

6:47 AM

Probably will be tomorrow

6:49 AM

Or you can try that https://github.com/Rain92/kobo-qt-setup-scripts But I didn't test it so I don't know what it does really

GitHub - Rain92/kobo-qt-setup-scripts

Contribute to Rain92/kobo-qt-setup-scripts development by creating an account on GitHub.

6:50 AM

And then compile InkBox yourself qmake . make

tux-linux

12:40AM here so

Diffrent time zones, a 5 hour shift

I'm in Canada ;p

3:31 PM

I'll make the GUI bundle now

Bundle done!

3:41 PM

http://23.163.0.39/bundles/private/update-n306.isa

Szybet pinned a message to this channel. 3/7/2022 4:10 PM

Szybet pinned a message to this channel. 3/7/2022 4:11 PM

i will try it

1

the linus book doesn't work

not again

4:21 PM

have you global reading settings enabled?

yes

hang on let me try here

4:22 PM

maybe it's an issue with the book itself

4:22 PM

do other ones work?

no, epubs dont work

linus book works perfectly here

4:24 PM

did you reboot?

yes, i can reboot one more time

4:25 PM

for me none epubs work

do reboot

4:25 PM

this is weird

oh, one works

is it the one not in a subdir for some reason?

all of them are in a subdir

do you open them with the recent books buttons or the file manager?

both

4:26 PM

even with search

does search work with subdirs?

yes

nice

4:27 PM

I can't see what's causing the issue

4:28 PM

while true; do ls -shal /kobo/run /kobo/inkbox; sleep 1; done

?

please run this on the kobo

4:28 PM

open an ePUB, then a PDF, then an ePUB

with what, search, file, recent?

file manager

does that make a diffrence when i use something else?

something else what?

4:29 PM

well, it should not

4:29 PM

(if you're talking about opening methods)

also, the file manager is slow, its hard to use so i avoid it as much i can

tux-linux

(if you're talking about opening methods)

yes

then Search is probably your best friend

4:30 PM

maybe I should just put a search-like file manager instead

no, a list is needed to

?

i dont remember names of all my books

No I mean, display a file list like if you searched for pdf, epub, png at once

4:32 PM

So it displays a list of all compatible books found

yes

4:32 PM

thats better

also, try to reboot

everything works now, i will reboot and use the last books menu

what?

4:34 PM

nonsense

4:34 PM

what did you do

yes it is

4:34 PM

the only thing i did in a other way, that i used the file manager not the search menu

wait; i'll try with the file manager on my side

4:34 PM

*search I mean

4:36 PM

oh; maybe the search function doesn't understand relative paths

4:36 PM

it doesn't work here too

4:36 PM

try putting linus book in onboard root and open it with search again

search doesn't work for me

4:38 PM

maybe add a check function that checks if a file exists

4:38 PM

or smth...

Szybet

search doesn't work for me

even if the file is in onboard root?

wait

4:39 PM

onboard works

ok, i see the problem

4:41 PM

https://github.com/Kobo-InkBox/inkbox/blob/8501b42124a8c44c79dfa455d3dfac577395bb99/generaldialog.cpp#L336-L357

4:41 PM

I only save dirIt.fileName() while I should also save dirIt.filePath() to allow opening files in subdirs

why does opening books take long, and next pages too? couldn't you save next pages into ram while reading current page?

I actually had an idea two days ago about this I'm going to implement it soon

good, then inkbox will be usable

Won't change the speed of opening a book much, but page will load in less than a second

Szybet

good, then inkbox will be usable

Did you try with smaller ebooks? It's faster there

4:44 PM

Search "jekyll" in online library and download the second book in the list

I opened a random book from recent list, it didn't worker ( last page, no previous page ), then the book dissapered from the list

4:46 PM

a book with only text works great

4:46 PM

but thats about it, some images and its dying

images will be something to implement too

why did the gui

4:47 PM

died

hmm hmm

and brightness to 0

ui freeze? (edited)

yes

4:47 PM

randomly

4:47 PM

i didn't do anything

Did you click "Cancel" in the file dialog?

4:48 PM

If you do ps | grep inkbox-bin, does that give anything?

im in the book that is fast

tux-linux

If you do ps | grep inkbox-bin, does that give anything?

yes 2203 root 0:09 /mnt/onboard/.adds/inkbox/inkbox-bin

can you try to click somewhere where it would normally do something?

nothing

then, put it to sleep; click the power button

and brighntess is 0

wait 15s click it again

4:49 PM

if you have access to the cli it is not dead

the brightness went to 100%, then slowly to 0 and now its sleeping

tux-linux

wait 15s click it again

After that, is it the same or can you click and it reacts?

but with no sleeping text

that's what I was fearing

4:50 PM

The Nia is now on the unstable devices list

4:50 PM

Try to wake it up

how did it launch a word seafrch list (edited)

4:50 PM

it waked up by his own, i didn't clicked anything

is your kobo plugged in via usb?

no

4:51 PM

i disconeccted it becouse everytime i go to home it asks me about usb

hmm

4:51 PM

can you try to put it to sleep one more time? try not to touch anything except the power button (edited)

its sleeps

well, yeah, but for how long

4:52 PM

the real moment when it sleeps is 15s after you pressed the power button

4:52 PM

otherwise it wakes up again because the kernel says "Resource busy"

it sleeps

ok oof

4:53 PM

now wake it up

4:53 PM

We will try something

and

4:54 PM

i cant click anything, the brightness button for example

4:54 PM

it went to a book

4:54 PM

why

4:54 PM

another freeze

cp /etc/init.d/inkbox_gui /tmp
vi /tmp/inkbox_gui

4:55 PM

Edit the file so that line 47 looks like this

4:55 PM

Then

mount --bind /tmp/inkbox_gui /etc/init.d/inkbox_gui
rc-service inkbox_gui restart

4:55 PM

This will launch InkBox GUI with the newer platform plugin I was talking about yesterday

4:57 PM

Also, when you run chroot /kobo /bin/kobo_config.sh it should output luna, right?

tux-linux

Also, when you run chroot /kobo /bin/kobo_config.sh it should output luna, right?

yes

ok, so try the steps above

4:59 PM

I'm trying them on my Libra too which has the same problem

i need wifi but i dont have any gui

don't you have serial?

do i need to try these 2 commands from yeastarday?

no

tux-linux

don't you have serial?

nano sucks on serial

use vi

5:00 PM

also, tip: reset so that your terminal isn't messed up anymore

5:01 PM

or maybe you can usbnet?

i will just do it with vim

5:01 PM

i will try

noooooo

5:01 PM

serial doesn't like vim

5:01 PM

only likes vi

yes vi vi

1

well, it messes up rotation on my libra

5:05 PM

hope it won't be the case for the Nia

[FBInk] Unidentified Kobo device code (40)!                                                    
kobofb: Finished initialization.                                                               
[FBInk] [refresh_kobo] MXCFB_SEND_UPDATE_V1_NTX: Operation not permitted!                      
[FBInk] [fbink_refresh] Failed to refresh the screen!                                          
[FBInk] [refresh_kobo] MXCFB_SEND_UPDATE_V1_NTX: Operation not permitted!                      
[FBInk] [fbink_refresh] Failed to refresh the screen!                                          
[FBInk] [refresh_kobo] MXCFB_SEND_UPDATE_V1_NTX: Operation not permitted!                      
[FBInk] [fbink_refresh] Failed to refresh the screen!

5:09 PM

mhm

so you can never see the gui?

the screen doesn't changed

cat /sys/class/graphics/fb0/rotate

3

fbink -k -f does something?

[FBInk] Couldn't find a Kobo version tag (onboard unmounted or not running on a Kobo?)!        
[FBInk] Detected a Kobo Nia (382 => Luna @ Mark 7)                                             
[FBInk] Enabled Kobo Mark 7 quirks                                                             
[FBInk] Clock tick frequency appears to be 100 Hz                                              
[FBInk] Screen density set to 212 dpi                                                          
[FBInk] Variable fb info: 758x1024, 8bpp @ rotation: 3 (Counter Clockwise, 270°)               
[FBInk] Fixed fb info: ID is "mxc_epdc_fb", length of fb mem: 3145728 bytes & line length: 768s
[FBInk] Canonical rotation: 0 (Upright, 0°)                                                    
[FBInk] Fontsize set to 16x16 (IBM base glyph size: 8x8)                                       
[FBInk] Line length: 47 cols, Page size: 64 rows                                               
[FBInk] Pen colors set to #000000 for the foreground and #FFFFFF for the background            
[FBInk] [refresh_kobo_mk7] MXCFB_SEND_UPDATE_V2: Operation not permitted!                      
[FBInk] [fbink_cls] Failed to refresh the screen!

ok, now try to put it to sleep, then wake it up maybe we can be lucky

5:14 PM

After that, try rc-service inkbox_gui restart

only the minutes showed up

5:15 PM

after returning from sleep

ok, yeah, good thing

5:15 PM

try service restart now

yes, its working

5:16 PM

whot now

well, what is working? the UI?

5:17 PM

touch input, graphics, all?

i managed to change the brightness, so yes

ok, rotation is good?

5:17 PM

for a minute it looked like this

5:17 PM

how to i rotate?

ls -shal /kobo/mnt/onboard/.adds/qt-linux-5.15.2-kobo/lib/plugins/platforms/libkobo.so

Szybet

for a minute it looked like this

The Glo HD does this too when getting out of sleep, idk why

tux-linux

ls -shal /kobo/mnt/onboard/.adds/qt-linux-5.15.2-kobo/lib/plugins/platforms/libkobo.so

no such file

it is fixed by restarting inkbox binary

Szybet

no such file

ls -shal /kobo/mnt/onboard/.adds/qt-linux-5.15.2-kobo/plugins/platforms/libkobo.so sorry

411 -rwxr-xr-x 1 root root 411.6K Feb 19 17:24 /kobo/mnt/onboard/.adds/qt-linux-5.15.2-kobo/plugins/platforms/libkobo.so

good!

5:22 PM

Can you play around the UI a bit? Stress it maybe too?

yes

If it doesn't crash, wheehee (edited)

last books doesn't work

5:24 PM

already crashed

5:24 PM

and is refreshing

awesome

[FBInk] Unidentified Kobo device code (40)!
kobofb: Finished initialization.

tux-linux

awesome

why

yeah that's normal, it says that even when it's working

5:24 PM

sarcasm

5:24 PM

well, if that platform plugin doesn't work, we're screwed

hm

At least it's fixable by going to sleep, then waking up

5:25 PM

not requiring an entire reboot

still not good

i know

5:26 PM

but this is out of my knowledge

5:26 PM

it should not even happen

5:26 PM

also that's wonderful news for clara hd porting (same kernel, or maybe andi1's 5.16 tree will save us) delivering an image that is half-working at best (edited)

5:29 PM

at what rate did it crash before we tried with the newer platform plugin?

5:29 PM

like, 5 mins average, 10 min average, what?

5:32 PM

I see one possible, hypothetic, crazy solution: You implement a VNC viewer with FBInk in Rust and we run that one instead of the native platform plugin. In Qt libs I put on the Kobo, there's libvnc.so, which allows creating a VNC server to which you can connect with a client. It is used in InkBox OS emulator to interact with the UI because implementing a touch screen in QEMU would be difficult. Assuming that your VNC viewer maps touch input correctly and does not trigger the same kernel bug that the platform plugin did, this would provide a complete, stable and painless UI hardware abstraction layer. (edited)

5:34 PM

Maybe you can look at this project https://github.com/whitequark/rust-vnc

tux-linux

I see one possible, hypothetic, crazy solution: You implement a VNC viewer with FBInk in Rust and we run that one instead of the native platform plugin. In Qt libs I put on the Kobo, there's libvnc.so, which allows creating a VNC server to which you can connect with a client. It is used in InkBox OS emulator to interact with the UI because implementing a touch screen in QEMU would be difficult. Assuming that your VNC viewer maps touch input correctly and does not trigger the same kernel bug that the platform plugin did, this would provide a complete, stable and painless UI hardware abstraction layer. (edited)

this will be not efficient, inkbox is now not super fast

then what do you suggest?

tux-linux

like, 5 mins average, 10 min average, what?

idk

tux-linux

then what do you suggest?

idk

5:38 PM

i dont know if it was that bad before

or we can just go ahead with what we have now

it wasn't that bad, you mean all crashes?

yeah

5:40 PM

ui freeze

some of them were becouse of the pdf - ebup thing

on my libra it is that bad

5:40 PM

maybe with some luck the Nia isn't as much as affected

maybe

5:44 PM

can you describe the issue more?

i can't, it's just as you said, it freezes randomly for no reason

5:45 PM

Qt still reacts to input events and button presses, it just doesn't show up on the screen

5:45 PM

killall inkbox-bin; env QT_QPA_PLATFORM=vnc:size=768x1024 chroot /kobo /mnt/onboard/.adds/inkbox/inkbox.sh You can try the VNC server this way

5:46 PM

Just access your Kobo's IP via Wi-Fi or USBNet with a VNC viewer (edited)

only qt?

5:46 PM

maybe its qt related

No, it happens when you do too much refreshes with FBInk too

why this problem doesn't occurs on the stock firmware?

no idea

maybe do a bug report to the linux kernel or smth?

what would that change?

5:51 PM

Check the average reply rate here: https://github.com/kobolabs/Kobo-Reader/issues (edited)

Issues · kobolabs/Kobo-Reader

Contribute to kobolabs/Kobo-Reader development by creating an account on GitHub.

or to the people that developed open source fbink?

Already asked; NiLuJe has no idea either

so hardware related?

maybe?

5:53 PM

probably

how to restore the changes? (edited)

5:56 PM

now nothing is working

Reboot

nope, i rebooted twice

Huh

5:59 PM

It should have unmounted the things

it is proprably software related? if hardware it would occur on stock firmware too

What's /etc/init.d/inkbox_gui contents?

6:01 PM

the changes are gone

6:01 PM

maybe i will dissconnect the battery?

It should work

6:04 PM

Not the battery

6:04 PM

I don't know why it doesn't

6:04 PM

It crashes or what?

flickers

6:04 PM

and nothing works

What do you mean "flickers"? How

the screen

Did you try service restart?

i rebooted twice...

Boot Nickel and come back

restore the sd card?

Don't you have another one?

no

Well

6:08 PM

Idk

6:08 PM

Do you have a video of what's happening?

dissconecting the battery worked

6:10 PM

its working like before (edited)

WHAT

6:13 PM

HOW

6:13 PM

huh ok

if you could repair the search menu / last books menu i will take it to school / use the ereader with inkbox regularly and check how much the problem occurs

1

Your Kobo is weird

tux-linux

HOW

Szybet

if you could repair the search menu / last books menu i will take it to school / use the ereader with inkbox regularly and check how much the problem occurs

I'll start with search

it would be awesome if you managed to do this today

6:15 PM

today in my time zone*

Maybe you can expect it by about 10pm in your timezone?

6:17 PM

Or maybe earlier

sure, till 11 pm i can apply the patches

ok, fixed the search issue

good, thats the most important now

8:09 PM

the last books menu is broken in many other ways, so well

8:09 PM

search the best

Szybet

the last books menu is broken in many other ways, so well

yeah, i know, I'm planning to remove that or replace it with something else

just a list with a slide, and maybe show directories with and then the books with a tab

https://github.com/Kobo-InkBox/inkbox/commit/847aa0051ee7a623fd3fcdc5d6848fe44c019a58

Search dialog: Allow opening books from subdirectories · Kobo-InkBo...

8:17 PM

I'll make you a GUI bundle

tux-linux

yeah, i know, I'm planning to remove that or replace it with something else

maybe replace the file explorer + last books into one?

maybe

8:20 PM

could be a nice thing for 2.0 release

8:23 PM

Done, uploaded the bundle

8:23 PM

same url

tux-linux pinned a message to this channel. 3/7/2022 8:24 PM

is there a easier way to update it than going into u-boot?

scp it to the Kobo via usbnet

8:26 PM

or make your own update package, but that's more complicated than going into u-boot

tux-linux

scp it to the Kobo via usbnet

thats too

that much?

8:27 PM

scp update-n306.isa root@192.168.2.2:/data/firmware/update.isa and you're done

8:27 PM

(well, service restart needed after that, but it's basically it)

tux-linux

that much?

maybe

that's what I do all the time TBH

Search works

nice

8:38 PM

and you can open books from a subdir?

yes

8:38 PM

even the linus epub with it

nice

8:39 PM

and you're able to switch between pdf<->epub no problem?

yes

Ok awesome!

i think battery life on inkbox will be better?

8:45 PM

with suspend?

yes

8:46 PM

Well, it should be

the diode is not on when charging?

9:08 PM

+ its sleeping

No

9:10 PM

In what way is this a problem?

9:10 PM

It is just handled by software and Nickel normally

i want to know if its charging

9:25 PM

thats another option to add

Szybet

i want to know if its charging

there's the ~ battery icon ~

its sleeping

9:31 PM

hmm

ah

9:35 PM

well, none of my kobos do that and I survive

i dont want to trust the usb charger if its charging

1

9:36 PM

it went out of sleep after hitting 100%

9:44 PM

bug or feature?

bug?

9:45 PM

idk why

9:45 PM

kernel-related

i forgot about one thing

10:08 PM

suspend when a magnet is near the hall sensor

10:09 PM

now its only getting it out of suspend, but not in

hmm, yeah, that's an input event code I forgot to add

10:11 PM

1. Reboot the Kobo. 2. When it says "(initrd) Press any key to stop auto-boot", press a key and login as root. 3. Launch evtest /dev/input/event0 4. Put the cover on the device as if you were putting it to sleep 5. Remove the cover from the screen 6. Post output of command in step 3.

can i do this using ssh?

you'll need to killall evtest before to ungrab the device i'd do it directly in the initrd

hm, okay

10:14 PM

but tommorow

1

A PDF opens, there is the first page but there are no othet pages to switch

The scale when changing pages is not applied, but the slider is still on its position (edited)

11:21 AM

Hm, ITS only in this specific pdf

11:24 AM

Also a random thought: why dont you use the library wiev for the last book section?

Szybet

Hm, ITS only in this specific pdf

When the PDF opens, its okay but aftet clicking options its doing this

The GUI freezes randomly, but not often. When i click power button its going to sleep, and the sleeping text appears. ( returning from sleep does not fix this ) Why? Its the most important problem @tux-linux (edited)

11:46 AM

It only flickers ( and freezes ) when i click some od the last books, its random (edited)

Szybet

The GUI freezes randomly, but not often. When i click power button its going to sleep, and the sleeping text appears. ( returning from sleep does not fix this ) Why? Its the most important problem @tux-linux (edited)

Oh, that's not what it does on my Libra Maybe it's a different problem On my Libra, if it freezes, you put it to sleep then wake it up and everything is fine

Szybet

Hm, ITS only in this specific pdf

Try with another PDF with a super long name

tux-linux

Try with another PDF with a super long name

HMMMMM okay

tux-linux

Oh, that's not what it does on my Libra Maybe it's a different problem On my Libra, if it freezes, you put it to sleep then wake it up and everything is fine

The screen works, but Qt maybe not

1:19 PM

It always freezes in a book in that problem

tux-linux

Try with another PDF with a super long name

Its not the case, the Linus one has a similar lenght

But it's an Epub

i dont have a long epub

3:37 PM

i will later change the name of one

Szybet

The GUI freezes randomly, but not often. When i click power button its going to sleep, and the sleeping text appears. ( returning from sleep does not fix this ) Why? Its the most important problem @tux-linux (edited)

how to debug this?

You can't

Szybet

i will later change the name of one

It's fixed, will commit shortly

tux-linux

You can't

what

tux-linux

You can't

Maybe check dmesg but don't be too hopeful

Transistor_Basics.pdf

499.89 KB

4:18 PM

does this pdf work for you?

https://inkbox.ddns.net/git/tux-linux/inkbox/-/commit/139aa787389fba9fbd2072f3609cc6ec39199d69

Reader: Set wordWrap on bookInfoLabel when book isn't an ePUB (139a...

Open-source eBook reader

4:28 PM

I'll see

im testing it for 4 hours now, changing pages, books, zooming etc

6:34 PM

everything works

6:34 PM

im confused

there are some problems with suspend

Szybet

im confused

Me too

Szybet

there are some problems with suspend

What exactly?

tux-linux

Me too

I will just test it till the weekend and then say

1

tux-linux

What exactly?

when it goes to suspend, after some seconds it goes back, but the screen is freezed, brightness on. To unfreeze it i need to put it to sleep once again. and this bug wasn't before, im confused. And wifi search after suspend doesn't work, it goes to a loop and i cant do nothing

8:40 PM

i will just test more

1

Szybet

when it goes to suspend, after some seconds it goes back, but the screen is freezed, brightness on. To unfreeze it i need to put it to sleep once again. and this bug wasn't before, im confused. And wifi search after suspend doesn't work, it goes to a loop and i cant do nothing

Wi-Fi in Kobos in general is not great from the start

the wifi chip in kobos is horrible. I have a similar one in a tablet and it works on linux without many problems from 4 months ( kernel 5.x )

8:42 PM

but it can be fixed, like a error message, not a loop

Wc121/cybertan is the worst ever.

also, here is the hal sensor

message.txt

4.98 KB

Oh thanks

1 is when the magnet is near, and 0 when it goes away

Ok yeah, so KEY_F1

8:46 PM

The interesting thing is when you run Nickel in an emulator pressing F1 on your keyboard triggers sleep mode

thats just like linux works

8:48 PM

windows key is alt + f1 for example

I honestly never use keybinds to put my computer to sleep mode

8:50 PM

So I learnt something new today

8:54 PM

I'm wondering if it would be possible to boot into the BDK on the Nia and other devices by taking the sleep cover or a magnet as the second button https://inkbox.ddns.net/wiki/index.php?title=Basic_diagnostics_kernel That would be nice

Basic diagnostics kernel

2 buttons

The Nia's U-Boot supports if/else statements, so maybe we could play around with that

Magnet + power button?

Yeah

8:56 PM

Or just the magnet Idk

Which what would that help, this kernel

Not much But it might be helpful for future similar devices which have an eMMC

I dont think anyone would want to risk bricking their ereader

Some did See Libra H2O porting thread

Nice

I have one, so maybe I'll try some things

Szybet pinned a message to this channel. 3/9/2022 1:03 PM

Is this usbnet static? Or do i nead to apply it every boot

It's static

echo true > /boot/flags/USBNET_ENABLE
echo 192.168.x.x > /boot/flags/USBNET_IP

and it will stay like that (edited)

3:18 PM

https://inkbox.ddns.net/wiki/index.php?title=Boot_flags Check docs if you want more info on boot flags

Boot flags

3:22 PM

There's something like ums in U-Boot you could do without using serial to launch it

3:22 PM

Direct firmware loader mode (edited)

3:23 PM

https://github.com/Kobo-InkBox/kernel/blob/master/initrd/common/rcS-root#L47-L64

kernel/rcS-root at master · Kobo-InkBox/kernel

Kernel sources for InkBox OS. Contribute to Kobo-InkBox/kernel development by creating an account on GitHub.

3:24 PM

And we could use KEY_F1 as the second key

3:27 PM

DFL basically exports the whole MMC over USB like U-Boot does but in Linux kernel

USBNET is everything i need

tux-linux

It's static

echo true > /boot/flags/USBNET_ENABLE
echo 192.168.x.x > /boot/flags/USBNET_IP

and it will stay like that (edited)

then there you go

changing a page boosts the cpu usage to 100%

6:52 PM

it goes to sleep, and then returns from it for no reason

7:32 PM

message.txt

50.51 KB

7:33 PM

here is part of dmesg

7:33 PM

maybe becouse of usbnet now?

it works after 5 times of trying, here is the dmesg for it. ,aybe there is some diffrence

message.txt

53.94 KB

1 Reboot 2 Try to suspend x times 3 after x times it works, and doesn't go back its random

8:26 PM

also, it doesn't go to sleep when wifi is connected, it goes back, always

8:27 PM

where is the script / command / code responsible for suspend? @tux-linux, i could propably debug this by myself (edited)

1 connect wifi 2 try to suspend 3 click wifi button 4 its freezed at searching for networks 5 press power button for sleep 6 it goes to sleep 7 go back 8 networks show up HMMM (edited)

when i turn on wifi, suspend doesn't work at all

8:57 PM

now i get it, The freezes in the GUI are propably always late suspend attempts

What's ls /dev/input?

12:40 AM

There should be event1 which should be the touchscreen

12:40 AM

Right?

8:51 AM

Yes, and?

8:54 AM

It is the touch screen, checked with cat

I leaved it on the home screen for -+ 1.5 hours, it eated up 3% od battery

1:02 PM

Still not bad

1:58 PM

Well, that's what I get there too

Szybet

Click to see attachment 🖼️

evtest /dev/input/event1 Please post event codes list

Szybet

changing a page boosts the cpu usage to 100%

but not for very long

Szybet

where is the script / command / code responsible for suspend? @tux-linux, i could propably debug this by myself (edited)

https://github.com/Kobo-InkBox/rootfs/blob/master/etc/init.d/sleep_standby.sh

rootfs/sleep_standby.sh at master · Kobo-InkBox/rootfs

InkBox OS root filesystem. Contribute to Kobo-InkBox/rootfs development by creating an account on GitHub.

tux-linux

evtest /dev/input/event1 Please post event codes list

Okay, later

3:43 PM

Why do you need it?

tux-linux

but not for very long

A patch is going?

Szybet

It is the touch screen, checked with cat

oh ok

Szybet

A patch is going?

will work on it this week-end probably

Szybet

A patch is going?

But "not for very long" meant it returns to idle after finished

3:44 PM

I lost 2% in 1h reading a book on the Glo HD, so it's not too bad

Szybet

Why do you need it?

I don't anymore But I'd like to see what mouse0 is up to evtest /dev/input/mouse0

tux-linux

But "not for very long" meant it returns to idle after finished

Yes, but its still not good

Well, I've been the only one programming this for an entire year... there are not-so-good things in InkBox, but people have to accept it I'll try to fix that issue this week-end

tux-linux

Well, I've been the only one programming this for an entire year... there are not-so-good things in InkBox, but people have to accept it I'll try to fix that issue this week-end

I will not accept it, i will patch it in the long future IF you dont

[FBInk] [refresh_kobo_mk7] MXCFB_SEND_UPDATE_V2: Operation not permitted!
[FBInk] [fbink_cls] Failed to refresh the screen!

now it started doing something like this EDIT: it happens only when KOBOX ( xorg ) is enabled interesting, right? do you know why? (edited)

what are those lines doing? https://github.com/Kobo-InkBox/rootfs/blob/5815843d6209188b8d07994bc47b3a1c3b85ca0b/etc/init.d/sleep_standby.sh#L15

im confused where is the part from going to sleep, that it returns to normal? i cant find it

tux-linux

I don't anymore But I'd like to see what mouse0 is up to evtest /dev/input/mouse0

reacts to touchscreen

what is this doing, why is nowhere something that returns this to a normal state

8:34 PM

if you would describe in short what does what it would be helpfull

[ 4673.449265] RTL871X: mmc1:0001:1: cannot remain alive while host is suspended
[ 4673.449290] dpm_run_callback(): pm_generic_suspend+0x0/0x30 returns -38
[ 4673.455921] PM: Device mmc1:0001:1 failed to suspend: error -38

hmm... (edited)

8:50 PM

kobo:/dev/input# echo "mem" > /sys/power/state
ash: write error: Function not implemented

8:51 PM

interesting too

Szybet pinned a message to this channel. 3/10/2022 9:10 PM

how to monitor the power button? / the signals that linux to sleep?

the git file: https://github.com/Kobo-InkBox/rootfs/blob/1129acbd9ae3018e4e93d0085512f89d342030bb/usr/local/bin/wifi/connect_to_network.sh and file on my device are diffrent, why is git not up to date?

rootfs/connect_to_network.sh at 1129acbd9ae3018e4e93d0085512f89d342...

InkBox OS root filesystem. Contribute to Kobo-InkBox/rootfs development by creating an account on GitHub.

9:37 PM

After rmmod modules from wifi, suspend works

I added this to /etc/init.d/sleep_standby.sh

if [ "$DEVICE" == "n306" ]; then
        killall -q dhcpcd wpa_supplicant
        ifconfig eth0 down
        rmmod 8189fs.ko 2> /dev/null
        rmmod sdio_wifi_pwr.ko 2> /dev/null
fi

and now suspend works without issues. I will test it more tommorow

only files associated with suspend? https://github.com/Kobo-InkBox/rootfs/blob/1129acbd9ae3018e4e93d0085512f89d342030bb/etc/init.d/idle_sleep https://github.com/Kobo-InkBox/rootfs/blob/1129acbd9ae3018e4e93d0085512f89d342030bb/etc/init.d/sleep_standby.sh#L18 https://github.com/Kobo-InkBox/rootfs/blob/1129acbd9ae3018e4e93d0085512f89d342030bb/etc/init.d/wake_standby.sh

Szybet

I added this to /etc/init.d/sleep_standby.sh

if [ "$DEVICE" == "n306" ]; then
        killall -q dhcpcd wpa_supplicant
        ifconfig eth0 down
        rmmod 8189fs.ko 2> /dev/null
        rmmod sdio_wifi_pwr.ko 2> /dev/null
fi

and now suspend works without issues. I will test it more tommorow

But then we have to handle to mess of reconnecting to the network after waking up ;p

Szybet

how to monitor the power button? / the signals that linux to sleep?

evtest /dev/input/event0

Szybet

what is this doing, why is nowhere something that returns this to a normal state

Basically, when it echo mem > /sys/power/state, the scripts hangs while the device sleeps, then when it wakes up, it executes the rest of the commands in it. (edited)

Szybet

[FBInk] [refresh_kobo_mk7] MXCFB_SEND_UPDATE_V2: Operation not permitted!
[FBInk] [fbink_cls] Failed to refresh the screen!

now it started doing something like this EDIT: it happens only when KOBOX ( xorg ) is enabled interesting, right? do you know why? (edited)

yes, very interesting, and no, I don't know why, although I'm sure X has something to do with it

Szybet

what are those lines doing? https://github.com/Kobo-InkBox/rootfs/blob/5815843d6209188b8d07994bc47b3a1c3b85ca0b/etc/init.d/sleep_standby.sh#L15

Emptying the /tmp/power file which contains event0 events to avoid memory leak (edited)

Szybet

the git file: https://github.com/Kobo-InkBox/rootfs/blob/1129acbd9ae3018e4e93d0085512f89d342030bb/usr/local/bin/wifi/connect_to_network.sh and file on my device are diffrent, why is git not up to date?

Maybe your rootfs' fuse-overlayfs write cache has something in it. Try:

mount /dev/mmcblk0p3 /mnt
rm -rf /mnt/rootfs/write
sync
umount /mnt
reboot

Szybet

only files associated with suspend? https://github.com/Kobo-InkBox/rootfs/blob/1129acbd9ae3018e4e93d0085512f89d342030bb/etc/init.d/idle_sleep https://github.com/Kobo-InkBox/rootfs/blob/1129acbd9ae3018e4e93d0085512f89d342030bb/etc/init.d/sleep_standby.sh#L18 https://github.com/Kobo-InkBox/rootfs/blob/1129acbd9ae3018e4e93d0085512f89d342030bb/etc/init.d/wake_standby.sh

Yeah, pretty much that (there are system services, like /etc/init.d/sleep_standby but they actually just launch the sleep_standby.sh script). (edited)

tux-linux

But then we have to handle to mess of reconnecting to the network after waking up ;p

The network works after that, it doesnt hang or freeze

tux-linux

evtest /dev/input/event0

And what file is reading from this? I want to debug why connecting or dosconnecting the cable gets the device from sleep (edited)

tux-linux

yes, very interesting, and no, I don't know why, although I'm sure X has something to do with it

Maybe ask the guy that developed it?

tux-linux

Maybe your rootfs' fuse-overlayfs write cache has something in it. Try:

mount /dev/mmcblk0p3 /mnt
rm -rf /mnt/rootfs/write
sync
umount /mnt
reboot

Hmm

tux-linux

Emptying the /tmp/power file which contains event0 events to avoid memory leak (edited)

Why it doesnt work in a test script on my PC? Diffrent sh versions?

7:34 AM

The GUI freezes and after 2 minutes this appears, closing it makes it work back (edited)

7:35 AM

7:37 AM

Also, the GUI freezed, and after a while ot jumped from page 2 to 44. Propably touch issues? Limit the page change or smth?

Szybet

Also, the GUI freezed, and after a while ot jumped from page 2 to 44. Propably touch issues? Limit the page change or smth?

It happens often

Szybet

The GUI freezes and after 2 minutes this appears, closing it makes it work back (edited)

This too

7:45 AM

Its nearly not usable like that (edited)

Do you know already that the epub viewer sometimes cuts off text when scalling? Even koreader has this problem, so i dont mind that much, but do you know or is it new

Szybet

Also, the GUI freezed, and after a while ot jumped from page 2 to 44. Propably touch issues? Limit the page change or smth?

Its horrible... Enable the next page button only the current site is loaded or something...

12:57 PM

️

Szybet

Why it doesnt work in a test script on my PC? Diffrent sh versions?

Because. No explanation to give. Things like that happen to me too and at that point I'm just like "fine"

Okay...

Szybet

Do you know already that the epub viewer sometimes cuts off text when scalling? Even koreader has this problem, so i dont mind that much, but do you know or is it new

Yeah, I don't have a fix, but you can use the scrollbars

Szybet

Its nearly not usable like that (edited)

So two days ago it worked no problem 4hrs and now it's nearly not usable? This thing is a nightmare to debug

Szybet

Its horrible... Enable the next page button only the current site is loaded or something...

Very good idea

tux-linux

So two days ago it worked no problem 4hrs and now it's nearly not usable? This thing is a nightmare to debug

It worked no problem becouse it was laying on a table, and i was clicking it carefully and slowly, but in real reading the problems accours

tux-linux

Yeah, I don't have a fix, but you can use the scrollbars

If there was a option to put the scrollballs even if the menu is off, it would make it the best PDF reader ( for pdfs that are images )

Szybet

If there was a option to put the scrollballs even if the menu is off, it would make it the best PDF reader ( for pdfs that are images )

There is

1:16 PM

Always show scrollbar In Settings

tux-linux

There is

Whete

1:16 PM

Really? Okay i will test it

1

1:17 PM

Suspend is working now, no problems after my fix, and the battery is holding great on suspend

1

Szybet

Click to see attachment 🖼️

Well, it's an English dictionary, so sorry

Szybet

The network works after that, it doesnt hang or freeze

It can't, you removed the Wi-Fi modules ! It's impossible !

tux-linux

Well, it's an English dictionary, so sorry

I mean, it freezes rabdomly and then the word search appears

Oh ok

tux-linux

It can't, you removed the Wi-Fi modules ! It's impossible !

The script is restoring them? Or is this sarcasm

No, it isn't, correct me if I'm wrong

I think it is (edited)

1:21 PM

There was insmod?

Can you check lsmod after waking up?

tux-linux

Can you check lsmod after waking up?

Not now, later

1

There doesn't seem to be anything like insmod there https://github.com/Kobo-InkBox/rootfs/blob/master/etc/init.d/wake_standby.sh

rootfs/wake_standby.sh at master · Kobo-InkBox/rootfs

InkBox OS root filesystem. Contribute to Kobo-InkBox/rootfs development by creating an account on GitHub.

Not this script

1:23 PM

This /usr/local/bin/wifi/connect_to_network.sh

Yeah but this isn't run when you get out of sleep

It is when i connect to wifi

Yes

1:25 PM

But if you rmmod Wi-Fi modules before going to sleep, it makes no sense that it works on wake up

OH, i need to recconect

1:25 PM

But thats normal?

Yes

1:26 PM

At least

1:26 PM

;p

For me it is, every device does that?

Since I don't unload the modules they stay connected

But the network dissconnects anyway?

I don't think so

Idk. Nia when goes to sleep dissconnects the usbnet, so Wi-Fi maybe too Wouldn't it better to unload modules for all devices? It saves battery

Yeah will do that

1:32 PM

And then reconnect with the script automatically

Okay, good

tux-linux

Yeah, I don't have a fix, but you can use the scrollbars

The scrollbar option fixes this

1:48 PM

The scrollbar if needed option is the best. Reading pdfs containing only images is a much better experience than in plato, koreader and stock nickel

1:48 PM

Fast, easy to use and clear. I would add it do the features od inkbox

1

tux-linux

Yeah will do that

Maybe also add updating the date and time from internet when connecting to a Wi-Fi. The rtc worka horrible

1

1:49 PM

I could add this too

Szybet pinned a message to this channel. 3/11/2022 1:58 PM

Well, not google.com but yea

Szybet

A patch is going?

Done

7:47 PM

Mainly here https://github.com/Kobo-InkBox/gui-bundle/commit/3e046379264009e83bfdab5a68f4862b821db91b

Greatly improve page turning speed · Kobo-InkBox/gui-bundle@3e04637

i can test it, if you create the gui bundle

in progress

7:49 PM

done

7:49 PM

same url

can i dirrectly update it from the kobo? i mean wget on the kobo to /opt/storage/update

yes

7:51 PM

but you need to specify -O

7:51 PM

Also, you need to rename update-n306.isa to update.isa

7:52 PM

So wget http://23.163.0.39/bundles/private/update-n306.isa -O /data/firmware/update.isa (edited)

data firmware?

Then sync and restart inkbox_gui

Szybet

data firmware?

It's a symlink that makes more sense than /opt/storage/update

well okay

thats fast and much better

Cool

8:00 PM

Now try to load a heavy ePUB, initial opening time will be longer but turning pages should be as fast (edited)

im trying to implement time sync on kobo, i figured out there is the ntpd command, but why ntpd -n -q -N -p time.cloudflare.com gives me a hour back than my real time?

tux-linux

Now try to load a heavy ePUB, initial opening time will be longer but turning pages should be as fast (edited)

only epub?

Yes

yea its faster

1

Szybet

im trying to implement time sync on kobo, i figured out there is the ntpd command, but why ntpd -n -q -N -p time.cloudflare.com gives me a hour back than my real time?

Maybe because you're syncing on UTC?

tux-linux

Maybe because you're syncing on UTC?

weird

8:06 PM

maybe wrong timezone?

Well I haven't implemented timezones in InkBox yet

8:06 PM

So it's on UTC

8:06 PM

It probably synced correctly

hm i see that https://time.is/pl/UTC

UTC - exact time now

Dokładny czas teraz, strefa czasowa, różnica czasu, czas wschodu/zachodu słońca i kluczowe fakty dla UTC.

8:07 PM

its 20:00 for me

1

tux-linux

Well I haven't implemented timezones in InkBox yet

do you plan to add it?

8:13 PM

I would like to add to /etc/ a file that says how much to add to UTC ( -1, +2 etc ) and a script in /opt/bin/ ( propably this patch, you can choose another ) that is runned when wifi connects, it uses /etc/ntp.conf for servers and sets the correct time.

8:14 PM

if you plan to add it in the future, you can build the timezones in gui and then change the file in /etc/ (edited)

Yes, that's about the way I was planning to do it

8:16 PM

ntpd never worked out for me for some reason but if you've got a command that works I'll take it

i can make an entire script, you just need to decide the syntax for /etc/utc_time.conf ( for example ) (edited)

Oh; I can do it, this was the second part of my goals for today

8:17 PM

But if you want to contribute you can do it

sure

But I was more thinking of something like /etc/localtime and symlinks

8:18 PM

But since the rootfs is r/o by default, it will be difficult

well, for this you need write privileges

And we can mount the rootfs r/w, but it is usually never done So I'd prefer not do it

8:19 PM

Maybe it could be possible to use bind mounts

maybe do tihs the same way you handle the config for the gui?

there is /usr/share/zoneinfo already there

8:20 PM

inkbox:/# ln -s /usr/share/zoneinfo/America/New_York /etc/localtime
inkbox:/# date
Fri Mar 11 14:20:25 EST 2022
inkbox:/#

Awesome

8:20 PM

It's the correct time here

i will try it for me

8:21 PM

yea. this is working

So maybe create /run/localtime or /tmp/localtime which is a symlink to /usr/share/zoneinfo/your/timezone and bind-mount that file to an empty file named localtime in /etc

does this need rw filesystem?

No

tux-linux

So maybe create /run/localtime or /tmp/localtime which is a symlink to /usr/share/zoneinfo/your/timezone and bind-mount that file to an empty file named localtime in /etc

um what

inkbox:/# ln -s /usr/share/zoneinfo/America/New_York /tmp/localtime
inkbox:/# date
Fri Mar 11 19:22:13 UTC 2022
inkbox:/# mount --bind /tmp/localtime /etc/localtime
inkbox:/# date
Fri Mar 11 14:22:26 EST 2022
inkbox:/#

This works

tux-linux

inkbox:/# ln -s /usr/share/zoneinfo/America/New_York /etc/localtime
inkbox:/# date
Fri Mar 11 14:20:25 EST 2022
inkbox:/#

Awesome

using this changed the time to the correct one, no need to move UTC hours (edited)

Yeah, so there's only ntp sync to implement really

give me 15 minutes

8:24 PM

is /usr/local/bin/wifi/connect_to_network.sh the main script when connecting to wifi?

Maybe a script that reads conf info from /data/config/19-timezone/config which contains timezone config, then syncs ntp and sets timezone in /etc/localtime

Szybet

is /usr/local/bin/wifi/connect_to_network.sh the main script when connecting to wifi?

I'd rather do that from the Qt code

8:25 PM

But one or the other works I guess

8:26 PM

Doing it from shell might be a better idea actually

Szybet

is /usr/local/bin/wifi/connect_to_network.sh the main script when connecting to wifi?

yeah, it is

tux-linux

yeah, it is

when its connected, i just need to add one line of code

Maybe launch /usr/local/bin/ntpsync.sh after udhcpc/dhcpcd complete

and one file to /etc

Yeah

8:26 PM

touch etc/localtime

tux-linux

Maybe launch /usr/local/bin/ntpsync.sh after udhcpc/dhcpcd complete

isin't needed

Szybet

isin't needed

Will be more portable

8:27 PM

If you want to run it from shell for example

8:27 PM

Maybe name it timesync.sh actually since it also changes the timezone

8:28 PM

Oh, never mind

look

8:28 PM

thats everything whots needed

There should be a system service that would change the timezone based on the config

why is the time in the GUI and cmd date diffrent...

Check /etc/init.d/hostname for example and create /etc/init.d/timezone which would change the timezone

tux-linux

Check /etc/init.d/hostname for example and create /etc/init.d/timezone which would change the timezone

first i woul need the config

Then chmod +x and symlink /etc/init.d/timezone to /etc/runlevels/boot/timezone to make it run at boot

8:31 PM

At /data/config? (edited)

8:31 PM

Yeah, I don't know how to handle that Also, the timezones list will be massive

8:32 PM

I'd suggest that /data/config/19-timezone/config would point to a path in the rootfs, for example /usr/share/zoneinfo/America/New_York

8:32 PM

Then the timezone service would symlink /tmp/localtime to the file specified in the config

tux-linux

Yeah, I don't know how to handle that Also, the timezones list will be massive

make a search bar

you implement this, not me.......

8:33 PM

this would probably be the most complicated thing ever

8:33 PM

well, idk

8:33 PM

depends on what we are talking about

8:34 PM

Maybe something similar to book search

tux-linux

inkbox:/# ln -s /usr/share/zoneinfo/America/New_York /tmp/localtime
inkbox:/# date
Fri Mar 11 19:22:13 UTC 2022
inkbox:/# mount --bind /tmp/localtime /etc/localtime
inkbox:/# date
Fri Mar 11 14:22:26 EST 2022
inkbox:/#

This works

does this ln -s dissapear after reboot?

yes

oh thats the problem

that's why we need to put it in a system service

8:34 PM

so it runs at each boot

8:34 PM

Actually, it might even be better if it was in the initrd

tux-linux

Actually, it might even be better if it was in the initrd

hm?

Well, some things like the ENCRYPT_LOCK flag which locks down the device for a day if the user enters too much wrong passphrases to unlock their encrypted storage would be affected

8:36 PM

But that's based on the epoch UNIX time

8:36 PM

So it might not be affected actually

8:36 PM

Ok, it won't be affected, I just checked

8:36 PM

So we can do this in the rootfs

if the file ends normarly without exit, then its connected? https://github.com/Kobo-InkBox/rootfs/blob/master/usr/local/bin/wifi/connect_to_network.sh#L78

yeah, otherwise the script calls exit 1 if there is an error

8:38 PM

like here https://github.com/Kobo-InkBox/rootfs/blob/5815843d6209188b8d07994bc47b3a1c3b85ca0b/usr/local/bin/wifi/connect_to_network.sh#L65

okay, so just add ntpd -n -q & at the end

8:39 PM

or without &

Not the &

8:39 PM

not at the end

yea, if the parrent exits it will kill him

8:40 PM

why?

maybe you can

why why not

i'm dumb https://github.com/Kobo-InkBox/rootfs/blob/5815843d6209188b8d07994bc47b3a1c3b85ca0b/usr/local/bin/wifi/connect_to_network.sh#L74

8:41 PM

$? is exit status of fi

8:41 PM

wow.....

O.o?

it will always quit on success if the dhcp request fails

well, then repair it

i'll make this a var real quick

you can add ntpd -n -q too

inkbox:/usr/share/openrc/support# test=1
inkbox:/usr/share/openrc/support# if test=1; then false; fi
inkbox:/usr/share/openrc/support# $?
-ash: 1: not found
inkbox:/usr/share/openrc/support# 

inkbox:/usr/share/openrc/support# if test=1; then true; fi
inkbox:/usr/share/openrc/support# $?
-ash: 0: not found
inkbox:/usr/share/openrc/support#

Actually, it seems I'm wrong (edited)

and /etc/ntp.conf server 0.pool.ntp.org server 1.pool.ntp.org server 2.pool.ntp.org server 3.pool.ntp.org (edited)

tux-linux

inkbox:/usr/share/openrc/support# test=1
inkbox:/usr/share/openrc/support# if test=1; then false; fi
inkbox:/usr/share/openrc/support# $?
-ash: 1: not found
inkbox:/usr/share/openrc/support# 

inkbox:/usr/share/openrc/support# if test=1; then true; fi
inkbox:/usr/share/openrc/support# $?
-ash: 0: not found
inkbox:/usr/share/openrc/support#

Actually, it seems I'm wrong (edited)

i dont understand at all

Szybet

i dont understand at all

Based on my logic, $? should always be 0 since the fi command is successful

8:45 PM

But it seems that ash is smart enough not to give fi an exit status

Szybet

why why not

You can add it at the end it's fine (edited)

wait

8:50 PM

couldn't we just symlink /etc/timezone to /data/config/19-timezone/config?

what?

or do you want in this config to write the path in text

Szybet

couldn't we just symlink /etc/timezone to /data/config/19-timezone/config?

/etc/localtime to /data/config/19-timezone/config? (edited)

Szybet

or do you want in this config to write the path in text

yes, or this?

/data/config/19-timezone/config would need to contain the contents of the corresponding file in /usr/share/zoneinfo

8:51 PM

or maybe make it a double symlink

tux-linux

/data/config/19-timezone/config would need to contain the contents of the corresponding file in /usr/share/zoneinfo

so a symlink could be good

i'm trying it now

and a system service wouldn't be needed

inkbox:/# ls -shal /etc/localtime
     0 lrwxrwxrwx    1 root     root          31 Mar 11 14:53 /etc/localtime -> /data/config/19-timezone/config
inkbox:/# ls -shal /data/config/19-timezone/config 
     0 lrwxrwxrwx    1 root     root          36 Mar 11 14:52 /data/config/19-timezone/config -> /usr/share/zoneinfo/America/New_York
inkbox:/# date
Fri Mar 11 14:53:59 EST 2022
inkbox:/#

It seems to work

8:54 PM

Go ahead

so a system service isin't needed

8:54 PM

great

ouch

inkbox:/# date
Fri Mar 11 14:53:30 EST 2022
inkbox:/# chroot /kobo
inkbox:/# date
Fri Mar 11 19:54:37 UTC 2022
inkbox:/#

8:55 PM

It seems it isn't working across chroots and /kobo is the GUI chroot

this explains why the date isin't correctly showing on the gui

tux-linux

It seems it isn't working across chroots and /kobo is the GUI chroot

maybe a reboot is needed? (edited)

Which, of course, doesn't contain zoneinfo (and the squashfs containing that filesystem isn't upgradable)

inkbox:/# ls usr/share/zoneinfo
ls: usr/share/zoneinfo: No such file or directory
inkbox:/#

(edited)

Szybet

maybe a reboot is needed? (edited)

i would doubt that

maybe copy the file to /kobo

8:57 PM

its only 3 KB

Maybe something similar to this [ ! -e "/kobo/opt/version" ] && touch /kobo/opt/version

8:59 PM

In that file https://github.com/Kobo-InkBox/rootfs/blob/master/etc/init.d/inkbox_gui

maybe a mount bind?

[ ! -e "/kobo/usr/share/zoneinfo" ] && mkdir -p /kobo/usr/share/zoneinfo
[ ! -e "/kobo/etc/localtime" ] && touch /kobo/etc/localtime
mount --bind /usr/share/zoneinfo /kobo/usr/share/zoneinfo
mount --bind /etc/localtime /kobo/etc/localtime

(edited)

exactly

I'll add that to the inkbox_gui, unless you want to do it

9:00 PM

Insert the above code under that line [ ! -e "/kobo/bin/kobo_config.sh" ] && touch /kobo/bin/kobo_config.sh

tux-linux

I'll add that to the inkbox_gui, unless you want to do it

you mean?

/etc/init.d/inkbox_gui

?

Since you are already working on other files, maybe it's good to have that in one commit only

message.txt

2.1 KB

9:02 PM

i edited the sleep_standby.sh to shut down wifi for all devices

9:02 PM

can you verify?

tux-linux

Since you are already working on other files, maybe it's good to have that in one commit only

on which files?

sleep_standby.sh for example Well, to avoid a merge by pulling my commits

tux-linux

sleep_standby.sh for example Well, to avoid a merge by pulling my commits

Its propaply the only one maybe connect_to_network.sh with the ntpd one line

Szybet

Click to see attachment 🖼️

It seems fine Now we need to tell it to auto-reconnect

and /etc/ntp.conf

tux-linux

It seems fine Now we need to tell it to auto-reconnect

wake standby script

yes

turning on the modules and interphace isin't enough?

is it enough on your kobo?

i can check that, but my kobo never recconects to a wifi

ok, I think I will do the modifications here and mention you in the commit

so you will push them all?

yeah

sure, it will be cleaner

send me the modified files (not just their contents, the file itself) and I'll put them in my tree

so: connect_to_network.sh /etc/ntp.conf wake standby sh sleep standby sh

9:10 PM

and there is the symlink of /etc/timezone to /data/config

/etc/localtime you mean?

yes

1

9:10 PM

wait, i will check it it recconects

1

Szybet

and there is the symlink of /etc/timezone to /data/config

i cant send it, so you need to symlink it on your own

sure

Szybet pinned a message to this channel. 3/11/2022 9:11 PM

% sshfs root@192.168.0.1:/ /mnt/Partition
root@192.168.0.1's password: 
read: Connection reset by peer

why sshfs is not working?

what are you trying to do?

9:13 PM

i think it needs a helper on the client

9:14 PM

did you modify the files directly on the kobo?

tux-linux

what are you trying to do?

sshfs is cool

i know

tux-linux

i think it needs a helper on the client

no, it workes on nickel

tux-linux

did you modify the files directly on the kobo?

i will scp

then no idea

Szybet

i will scp

mount /dev/mmcblk0p3 /mnt
mksquashfs /mnt/rootfs/write /tmp/rootfs-write.sqsh
umount /mnt

9:15 PM

That will take all the files you modified

why not simply ifsctl mnt rootfs rw

that mounts the rootfs rw

9:16 PM

what I wrote above takes all the write cache and compresses it into a squashfs

9:17 PM

so if you modified /usr/local/bin/wifi/connect_to_wifi.sh, /mnt/rootfs/write will contain a copy of this file that is modified it's why it's called overlayfs

oh okay

it overlays a writeable part over a read-only filesystem (squashfs)

Szybet pinned a message to this channel. 3/11/2022 9:17 PM

cool, but i already did ifsctl

it has nothing to do with that

9:18 PM

when you issue ifsctl mnt rootfs rw, it remounts the rootfs rw and all the writes are redirected to [p3]/rootfs/write

9:18 PM

so if you do touch /test, a file named test will be at [p3]/rootfs/write/test

9:18 PM

and the rootfs squashfs will stay the same

okay

that's why earlier you were wondering why your connect_to_wifi script didn't have the contents of the latest commits; because you had modified it before

9:19 PM

the modified version takes precedence on the existing version

oh...

9:21 PM

i forgot about a line in connect_to_network.sh, good that im testing that

1

have you tested if it reconnects automatically on wake up?

it does not even go to sleep, i need to add killall -q dhcpcd wpa_supplicant

ok

well, if i kill them and then launch there is no way they will reconect

before ifconfig wifi_dev down i guess?

yes

Szybet

well, if i kill them and then launch there is no way they will reconect

Yeah, so then we can work with the available info at /data/config/17-wifi_connection_information

9:24 PM

And re-launch /usr/local/bin/wifi/connect_to_network.sh with those strings (edited)

oh i can do this

it will mess up the tree

in a seperate file? like recconect.sh?

tux-linux

it will mess up the tree

you mean git? i can send you the file

yeah ok

9:25 PM

Well, I'll send you my sleep_standby.sh script before

thats mine

message.txt

2.42 KB

sleep_standby.sh

2.23 KB

they are the same, i used the one from git

I added a check to see if the interface is up or not before removing modules, etc.

oh okay

Szybet

they are the same, i used the one from git

no, it's the one with my additions + the wifi down part you added earlier

what about

if [ "${DEVICE}" == "n705" ] || [ "${DEVICE}" == "n905b" ] || [ "${DEVICE}" == "n905c" ] || [ "${DEVICE}" == "n613" ] || [ "${DEVICE}" == "n437" ]; then
    wlarm_le down
fi

oh yes

9:29 PM

added

sleep_standby.sh

2.4 KB

how to check if to restore wifi, when going from sleep?

Szybet

how to check if to restore wifi, when going from sleep?

if [ "$(cat /sys/class/net/${WIFI_DEV}/operstate)" == "up" ]; then
  echo "true" > /tmp/was_connected_to_wifi
fi

(edited)

9:33 PM

And deal with that file on wake up Don't forget to rm -f it once you've read its contents (edited)

9:36 PM

It's actually /sys/class/net/${WIFI_DEV}/operstate); added it above

if [ "$(cat /tmp/was_connected_to_wifi)" == "true" ]; then
    ifconfig "${WIFI_DEV}" up
    insmod "${WIFI_MODULE}" 2> /dev/null
    insmod "${SDIO_WIFI_PWR_MODULE}" 2> /dev/null
    ESSID=`cat /data/config/17-wifi_connection_information/essid`
    PASSPHRASE=`cat /data/config/17-wifi_connection_information/passphrase`
    /usr/local/bin/wifi/connect_to_network.sh "${ESSID}"  "${PASSPHRASE}" 
    rm -f /tmp/was_connected_to_wifi
fi

9:38 PM

i will test it now

better to use $(command) in place of backticks and if grep -q "true" /tmp/was_connected_to_wifi

9:39 PM

also, it's in the wrong order

tux-linux

also, it's in the wrong order

:/...

9:40 PM

okay

insmod sdio_wifi_pwr, then wifi module, then wlarm_le up for devices that need it, then interface up, then connect (edited)

oh right

9:41 PM

do i need it to change it to the grep?

if you want I can do it here after

9:41 PM

the important thing would be to get the snippet of code working

can you look at the "word search" and "pages are switching like crazy" problems i described before?

currently doing timezone stuff

okay

did it happen with the latest gui bundle? maybe the speed of the page turning has improved it

hmm, i will check it

9:45 PM

but a limiter for the page change is needed anyway

hmm

it works

nice

but it takes a while

how much time does it take?

9:52 PM

ah

idk, 90 s?

we can probably run it in the background

you mean

9:52 PM

?

if the user tries to connect to wifi in the gui during that period, we can show a message like "connection already in progress" or something

yes you can, a simple ps | grep connect_to_network could make it work

i was thinking of something more reliable than that ;p

9:54 PM

but yeah, it's the idea

9:55 PM

FWIW, since the GUI is running in a chroot, it needs to send commands to a fifo to get them executed in the rootfs That's how Wi-Fi connection is handled: https://github.com/Kobo-InkBox/rootfs/blob/5815843d6209188b8d07994bc47b3a1c3b85ca0b/etc/init.d/ibxd#L67-L77

complicated

yeah, but otherwise it's really a mess

9:56 PM

so the GUI has a while true loop which checks the contents of the wifi_connected_successfully file and stops when there is something like 'false' or 'true' in it

9:58 PM

yay

inkbox:~# date
Fri Mar 11 20:58:26 UTC 2022
inkbox:~# chroot /kobo
inkbox:/# date
Fri Mar 11 15:58:29 EST 2022
inkbox:/#

9:58 PM

(well, not the first part, but at least it works in the chroot)

syncing time also works

cool

so i will send the files

1

10:02 PM

ntp.conf

88 bytes

wake_standby.sh

2.59 KB

sleep_standby.sh

2.45 KB

connect_to_network.sh

2.08 KB

ok thanks

tux-linux

(well, not the first part, but at least it works in the chroot)

so the gui will show the right time?

yes

nice

I wonder if it should be /tmp/was_connected_to_wifi or /run/was_connected_to_wifi

idk, you decide

well, ibxd puts wifi_connected_successfully in run so I guess it'll be run

can i disable global readings settings for remembering when i last readed the book?

yes

10:08 PM

i worked on that this morning too it should be more reliable

10:09 PM

my reader framework is really a mess it took me 15 mins to understand what my own program is actually doing ;p (edited)

i think its normal

have you read reader.cpp ?

10:10 PM

https://github.com/Kobo-InkBox/inkbox/blob/master/reader.cpp

10:10 PM

good luck

10:10 PM

not everything is a mess in the file, but more the setup_book function and relatives

the change page button still does change too many pages when i click on it too much, but it does not freeze. ( so it will propably change to many pages when i will be using the ereader serious, in school or smth )

1

10:12 PM

so this needs to improve, and maybe thats it?

ok, so I'll try to implement some sort of 'blocker' while it's turning the pages

10:13 PM

this is fine, but that means it won't be able to go to sleep until it has reconnected to wi-fi

10:13 PM

10:14 PM

Since it's sleep_standby's job

maybe a button in settings to choose if it would recconect?

yeah, but if you move around and don't have access to the network anymore

10:15 PM

maybe a 'cancel reconnect' or something

is it normal that connecting the usb cable gets the device back from sleep?

Maybe the user would click one time to connect to Wi-Fi after waking up, it would say "Click again to cancel reconnect" and on the second click on the Wi-Fi icon it would cancel

Szybet

is it normal that connecting the usb cable gets the device back from sleep?

yes

10:16 PM

does that even in Nickel

tux-linux

yes

and when battery hits 100%?

no

tux-linux

Maybe the user would click one time to connect to Wi-Fi after waking up, it would say "Click again to cancel reconnect" and on the second click on the Wi-Fi icon it would cancel

maybe

1

but i don't really know what to tell you about this

its not a big issue

10:22 PM

did you implement the magnes thing?

Those are all my changes currently

rootfs.diff

8.19 KB

Szybet

did you implement the magnes thing?

sleep cover? no

tux-linux

Those are all my changes currently

I'll push a commit and you can try on your device

oh yes, i forgot about hwclock --systohc

done

tux-linux

sleep cover? no

could you? or is it a big issue to make

I guess it's easy

thats the last thing thats missing on nia

I forgot to mention you in the commit; I'll do it in the next one when we fix things I guess

maybe xorg, but xorg isin't important

Szybet

maybe xorg, but xorg isin't important

did you asked the maintainer?

no

10:27 PM

I highly doubt he has an idea

10:27 PM

The problem was that you couldn't suspend while in X, right?

Szybet

[FBInk] [refresh_kobo_mk7] MXCFB_SEND_UPDATE_V2: Operation not permitted!
[FBInk] [fbink_cls] Failed to refresh the screen!

now it started doing something like this EDIT: it happens only when KOBOX ( xorg ) is enabled interesting, right? do you know why? (edited)

this

10:28 PM

i think suspend is fixed for good, becouse of wifi

Szybet

[FBInk] [refresh_kobo_mk7] MXCFB_SEND_UPDATE_V2: Operation not permitted!
[FBInk] [fbink_cls] Failed to refresh the screen!

now it started doing something like this EDIT: it happens only when KOBOX ( xorg ) is enabled interesting, right? do you know why? (edited)

you mean, in Qt gui too?

well, in qt gui only?

so that means you have the problem I have on my Libra but I never disable X on my Libra so I think it is permanent

10:29 PM

oh; awesome

propably

well, not awesome, but it's nice to know that it can probably be circumvented on the libra by disabling x11

10:30 PM

and it does that randomly, right?

yes

10:30 PM

xorg isin't that important, those are ereaders

hmm

idle sleep is working too btw

10:31 PM

so everything, exept the magnet

nice

when will you implement the magnet?

this week-end?

you mean tommorow, sunday?

might be there

okay

10:33 PM

i will test more until then

actually it might just be as easy as that (edited)

rootfs.diff

414 bytes

10:34 PM

try it

its working

10:43 PM

its not a ideal and clear solution

10:44 PM

but it is

Szybet

its not a ideal and clear solution

?

/tmp/power looks like this most of the time:

Event: time 1647034909.855898, type 1 (EV_KEY), code 59 (KEY_F1), value 0
Event: time 1647034909.855898, -------------- SYN_REPORT ------------

yeah

10:45 PM

and?

its writing to it all the time

10:45 PM

is it good? performance eating?

no

10:45 PM

usually, CPU is at 2-3% at idle

its jumping 3-6-8-12% all the time

10:47 PM

not stable

10:47 PM

its not bad

1

10:47 PM

and the cpu is propably weakier

10:50 PM

is the time in the gui fixed, all commits pushed? i would like to update the rootfs

not fixed yet

10:51 PM

but you can update the rootfs

okay

tux-linux

not fixed yet

well, you can't set the timezone in the gui

10:51 PM

if you set via cli, it works

thats what i asked for

erase your overlayfs cache

tux-linux pinned a message to this channel. 3/11/2022 10:52 PM

I pinned the commands

hmm scp rootfs.squashfs root@192.168.0.1:/mnt/ now i know this isin't a good idea

11:10 PM

so how do i update the rootfs without u-boot?

you technically can but load busybox binary into ram before (just run busybox) without any arguments then busybox sync busybox reboot (edited)

11:16 PM

it's not clean, it's an horrible idea, but I guess it does the job

nope, nothing worked

11:17 PM

i will tommorow open in and do it properly

11:17 PM

what is the best way to do this ( without opening )? (edited)

Szybet

i will tommorow open in and do it properly

did it transfer?

Szybet

what is the best way to do this ( without opening )? (edited)

via u-boot/DFL or an update package (edited)

tux-linux

did it transfer?

it did ( i think ) , then everything stopped working and i couldn't the openssl file

you should have scp'ed them at the same time

true

11:20 PM

after error code 3, does it reboot or go shutdown?

shutdown

nice

https://github.com/Kobo-InkBox/kernel/blob/439f7b42a06371d473995eab3fa43d88f7779498/initrd/common/overlay-mount#L41-L49

https://github.com/Kobo-InkBox/rootfs/commit/b20aaad03c2b44ccca276512fa1fed039a8dcd27

Include sleep cover in sleep mode triggers · Kobo-InkBox/rootfs@b20...

Thanks to @Szybet for the contributions in this commit and in 3fb505c19b03bb8f164c185b86b86304c4ff7d75

Done! https://github.com/Kobo-InkBox/rootfs/commit/0024535e5ce23b3ad6a97669ce90e009dd0fc6fb https://github.com/Kobo-InkBox/inkbox/commit/1aca2ba6156035b540c529dea0cd8930f2c68c2a

Wi-Fi: Disable when going to sleep and auto-connect on wake-up · Ko...

Wi-Fi: Auto-connect on wake-up things · Kobo-InkBox/inkbox@1aca2ba

https://github.com/Kobo-InkBox/inkbox/commit/f0c02a37d6cfcd6cdc8426ce78f69bd0f359289f

Prevent massive page turning hang · Kobo-InkBox/inkbox@f0c02a3

You can try encrypted storage if you want btw

tux-linux

Done! https://github.com/Kobo-InkBox/rootfs/commit/0024535e5ce23b3ad6a97669ce90e009dd0fc6fb https://github.com/Kobo-InkBox/inkbox/commit/1aca2ba6156035b540c529dea0cd8930f2c68c2a

I updated the GUI bundle at the usual URL (edited)

6:57 AM

So you need to update the rootfs and the GUI bundle

with what do you open your kobos? im using this below but it damages a bit the plastic. maybe there is a better way

credit card

no way ot will work (edited)

1:57 PM

maybe the nia is tightener

usually I start if a knife, and once some place in the back has enough damage I can fit a credit card to open the entire device

1:57 PM

well, it isn't ideal, but that's what happened with my Libra

a knife will damage it too

there's something called a spudger

1:58 PM

i don't have that

1:58 PM

this one?

more like that

tryied with it, no chance

2:00 PM

( for the nia )

well, then I don't know

2:12 PM

this e

(edited)

Whoa

2:17 PM

It should not be like that

2:17 PM

Do you have x enabled?

tux-linux

https://github.com/Kobo-InkBox/inkbox/commit/f0c02a37d6cfcd6cdc8426ce78f69bd0f359289f

i dont really know if this patched the issue

tux-linux

Do you have x enabled?

disabled

Szybet

i dont really know if this patched the issue

Well, it disables the button when you click it and enables it back when the page is loaded

2:21 PM

If that doesn't fix your issue then nothing can

tux-linux

Well, it disables the button when you click it and enables it back when the page is loaded

so in theory if i spam click the button it shouldn't switch more than 1 page?

No It will switch one, two, three pages once the page before has been correctly loaded, although might be not fully printed on the screen yet

2:22 PM

So it ensures each page is properly loaded before loading a new one

so it doesn't work

Well, use the go to function if you want to spam page turns

i mean that propably when using the reader some times it will jump 2 pages

It never does that on my devices

maybe i click not exact

2:26 PM

its not a big issue

Ok, what about Wi-Fi reconnection on wake up?

tux-linux

Ok, what about Wi-Fi reconnection on wake up?

i will check it, first how do i set the timezone now

2:27 PM

from cli

You need to symlink a zoneinfo file to /data/config/19-timezone/config Use ln -sf

ln -sf /data/config/19-timezone/config /usr/share/zoneinfo/Poland ?

The opposite

2:29 PM

But yeah

ln -sf /usr/share/zoneinfo/Poland /data/config/19-timezone/config

2:30 PM

okay

Yes

recconecting works

2:42 PM

time works too

2:42 PM

perfect

Awesome

Szybet

Click to see attachment 🖼️

um, when did you fixed this

2:48 PM

it works

2:49 PM

where is the list of supported devices?

2:51 PM

here i see it: https://inkbox.ddns.net/wiki/index.php?title=Main_Page

Main Page

2:52 PM

i really think you can add to them the kobo nia

2:52 PM

and the kobo libra, with a note that xorg doesn't work correctly

I'll add them once I release 1.9

sure

3:03 PM

I have closed the issue on github

tux-linux

Ok, what about Wi-Fi reconnection on wake up?

i cant go to sleep until it recconects, but yea

3:16 PM

oh, i need to click 2 times

3:16 PM

okay

3:17 PM

and it went back from sleep and asks me to choose the network

Yeah, because you clicked two times

i have updated the guide, some parts need more clarification https://github.com/Szybet/kobo-nia-audio/blob/main/inkbox.md

kobo-nia-audio/inkbox.md at main · Szybet/kobo-nia-audio

Kobo nia ereader with audio added. Contribute to Szybet/kobo-nia-audio development by creating an account on GitHub.

1

idk why, here is the fix: https://github.com/Kobo-InkBox/rootfs/pull/1

fix sometimes appearing sleeping issues on the kobo nia by Szybet ·...

The device will sometimes not go to sleep when the modules are loaded, even if the interface is down

No, it's because the operstate file in sysfs shows down even if the modules are loaded and the interface is up, it shows "up" only when the interface has got an IP

its still if its down

Szybet

its still if its down

?

tux-linux

No, it's because the operstate file in sysfs shows down even if the modules are loaded and the interface is up, it shows "up" only when the interface has got an IP

yes, and?

Well, if you searched for networks but didn't connect, it will fail to sleep

look 1 interface is down 2 modules are loaded 3 sleep doesn't work

Yes

12:33 AM

Idk

"tap again" where to tap? at the wifi button?

Maybe get out the block of code removing the WiFi modules from the if statement and put it before

Szybet

"tap again" where to tap? at the wifi button?

Yes? Isn't that intuitive?

tux-linux

Yes? Isn't that intuitive?

i ask becouse it doesn't work. when i click again it shows the "searching for network" label

Yes, that's what it's supposed to do ...?

tux-linux

Maybe get out the block of code removing the WiFi modules from the if statement and put it before

then it will rm them 2 times sometimes. But it doesn't matter, do as you would like, its just code preferency look

Szybet

then it will rm them 2 times sometimes. But it doesn't matter, do as you would like, its just code preferency look

I'll do that later tonight

tux-linux

Yes, that's what it's supposed to do ...?

Tap again to cancel, not to search for networks?

To cancel the reconnection

12:37 AM

And connect to another network

12:38 AM

That's how I see it ?...

A

12:38 AM

o thought it will shut down the wifi

1

12:44 AM

could i add a service that will check the power button, and if it will clicked it will stop wake_standby and start sleep_standby? it would remove the problem that after going from sleep it cant go to sleep again? (edited)

That would cause some unwanted problems

Maybe I could add something related to what Qt does when it cancels reconnect

Szybet

could i add a service that will check the power button, and if it will clicked it will stop wake_standby and start sleep_standby? it would remove the problem that after going from sleep it cant go to sleep again? (edited)

https://github.com/Kobo-InkBox/rootfs/commit/6eb0ad2f626556c0322c178fbadd47b5aac0632c This was a nightmare.

Wi-Fi auto-connect on wake-up: lot of bugfixes · Kobo-InkBox/rootfs...

7:01 AM

But at least it's done and it works (hopefully) Edit: don't expect it to work. I tested on my Touch and it's still not acting properly. Will try to fix soon (edited)

But if you can, please look and tell me if you see something in the commit that should not be there or if it is missing. The modules removal part is 100% fixed now, but going to sleep while reconnecting still gives problems.

7:28 AM

I'm going to sleep now. Enough of this shit for tonight

https://github.com/Kobo-InkBox/rootfs/commit/a3e96f2a0bf6a964967a98acb7f0280c29d67849 It seems to be better now

Even more bugfixes for Wi-Fi reconnection on wake-up · Kobo-InkBox/...

Its even more complicated

i know

2:30 PM

but it works on my Kobo, so I guess it might on yours too

i will test later could you show me a place where i can ask question about qt? i have some questions and i cant find answers

Qt discord

2:36 PM

But it might be too specific if you want to setup a cross compile environment; maybe ask pgaskin, NiLuJe or sherm_p on this server

tux-linux

Qt discord

i will try

tux-linux

But it might be too specific if you want to setup a cross compile environment; maybe ask pgaskin, NiLuJe or sherm_p on this server

first I want to learn qt, then use it on the kobo

1

https://github.com/Kobo-InkBox/inkbox/commit/62b09f40624ac94ecdc9f2b36b94d90844915792

Settings: Add an option to change timezone · Kobo-InkBox/inkbox@62b...

1

tux-linux

but it works on my Kobo, so I guess it might on yours too

what should i test?

7:22 PM

also, how bad is porting a qt cpp app to kobo?

Szybet

also, how bad is porting a qt cpp app to kobo?

once you have the setup, it's easy

Szybet

what should i test?

1. Put it to sleep and wake it up, not connected to any Wi-Fi network. Should work as usual. 2. Connect to a Wi-Fi network. Then, put it to sleep and wake it up. 3. Wait until the Wi-Fi has reconnected (the Wi-Fi icon updates each 10s) and then put it to sleep. 4. Wake it up and put it back to sleep again while it's reconnecting. Then wake it up. It should not attempt to connect to any network at this point.

4 doesn't work

which part?

4.

Well, it fails at which sub-step?

you know that nia will not go to sleep if modules are loaded?

yes

7:38 PM

They are removed

7:39 PM

https://github.com/Kobo-InkBox/rootfs/blob/a3e96f2a0bf6a964967a98acb7f0280c29d67849/etc/init.d/sleep_standby.sh#L60-L91 Even if the interface is down, they will be removed

it simply doesn't respond for the power button

7:40 PM

no sleeping splash screen

can you still interact with the ui? (edited)

Szybet

it simply doesn't respond for the power button

how much time did you wait after it woke up and started reconnecting?

tux-linux

how much time did you wait after it woke up and started reconnecting?

30s

Maybe it had already reconnected (edited)

i mean

7:42 PM

after going back from sleep i clicked the power button instantly

7:42 PM

but no reaction

Szybet

but no reaction

I think it's at least 3-4 secs before it can react, it has to go through other steps in the scripts before

7:47 PM

I'll try on other models

tux-linux

I think it's at least 3-4 secs before it can react, it has to go through other steps in the scripts before

nope, still doesn't react

Ok. Please reboot, then do the following:

cp /etc/init.d/sleep_standby.sh /etc/init.d/wake_standby.sh /tmp
vi /tmp/sleep_standby.sh # Replace first line with '#!/bin/bash -x'
vi /tmp/wake_standby.sh # Replace first line with '#!/bin/bash -x'
mount --bind /tmp/sleep_standby.sh /etc/init.d/sleep_standby.sh
mount --bind /tmp/wake_standby.sh /etc/init.d/wake_standby.sh
killall sleep_standby.sh wake_standby.sh # If nothing found, try to find processes with `ps' and kill the PID related to those scripts
rc-service sleep_standby stop
rc-service wake_standby stop
rc-service sleep_standby start -d

Then redo the steps above and post terminal output. It's a truckload of info, so it will probably be useful for debugging.

7:53 PM

Actually, only do step 3 and 4.

Szybet

nope, still doesn't react

have you tried pressing it more than one time?

tux-linux

have you tried pressing it more than one time?

yes, and then after 2 minutes it gos to sleep and doesn't want to go back

well, that is a kernel issue no idea

7:58 PM

The early wake-up part is handled by the kernel If the screen didn't refresh when you clicked the power button, it's a kernel issue

it works, i needed to clear the write cache something

8:33 PM

Szybet

it works, i needed to clear the write cache something

What works?

everything

tux-linux

1. Put it to sleep and wake it up, not connected to any Wi-Fi network. Should work as usual. 2. Connect to a Wi-Fi network. Then, put it to sleep and wake it up. 3. Wait until the Wi-Fi has reconnected (the Wi-Fi icon updates each 10s) and then put it to sleep. 4. Wake it up and put it back to sleep again while it's reconnecting. Then wake it up. It should not attempt to connect to any network at this point.

from here

Nice

usb storage will not work with usbnet, right?

yes, but it removes usbnet modules before inserting mass storage ones

10:46 PM

so you lose usbnet while in mass storage

10:46 PM

it restarts it after if you have it enabled

so nothing works?

not while you're in mass storage

10:47 PM

wi-fi still works though

10:47 PM

you can ssh that way

sshfs would be nice to get working

you can try

to enable sshfs its needed to add Subsystem sftp internal-sftp to /etc/ssh/sshd_config

Szybet pinned a message to this channel. 3/13/2022 11:03 PM

Szybet

to enable sshfs its needed to add Subsystem sftp internal-sftp to /etc/ssh/sshd_config

oh cool

writing doesn't work

oh?

i will debug it later

11:08 PM

using sshfs

did you mount rootfs r/w? (edited)

/data/onboard/

strange

scp works

11:08 PM

then idk

I put yesterday my nia in the bag, now I look it was in the menu and 39%... Weird

3:22 PM

I have a case that prevents the power key from being clicked, and it goes to sleep after 15 minutes

idk

6:38 PM

Sleep timeout is a nice workaround

tux-linux

Sleep timeout is a nice workaround

I had it enabled

6:41 PM

strange

6:42 PM

i will test it more, maybe a one time event ( maybe the ereader was next to a tablet, and the tablet had magnets in it )

i charged it up at 17:00, now at 21:30 it has 87%

9:28 PM

thats concerning

9:29 PM

and it was sleeping the whole time

9:29 PM

maybe becouse of usbnet?

Strange

there is no way powertop would work there, right?

Idk

9:41 PM

You can check cpu usage with htop maybe

it jumps 4-8-12% all the tim

Maybe check dmesg, there might be power info

message.txt

241.92 KB

9:44 PM

[    5.614571] syscon-poweroff 20cc000.snvs:snvs-poweroff: pm_power_off already claimed 80021c08 ntx_machine_poweroff
[    5.625116] syscon-poweroff: probe of 20cc000.snvs:snvs-poweroff failed with error -16

9:46 PM

maybe some kernel options to enable more power saving features?

9:46 PM

i will need to tinker with that

I'm planning to release 1.9 soon, since we have pretty much everything ready. Any features/fixes/something I forgot about?

sliders too small

+ settings arrows to choose numbers are to small

last book menu is simply not good
improve wifi control maybe?
file chooser those are the most important in my opinion

(edited)

those will be for 2.0

10:20 PM

Maybe the last thing to add to 1.9 would be an option to turn wi-fi off

yea

i don't know if something is possible for the sliders and settings arrows

10:21 PM

it's like we'd have to scale only one part of the UI but not all of it idk

settings arrows - replace with something else for sliders it must be possible, QT is not that static?

Szybet

settings arrows - replace with something else for sliders it must be possible, QT is not that static?

what? it's a qspinbox for sliders... maybe (edited)

https://stackoverflow.com/questions/47691972/qslider-increase-handle-size random answer from internet

QSlider increase handle size

I need badly to increase size of handle for my slider, but no css options can do this(styleSheet()). The default example from Qt docs didn't help me as well. I have a slider like this: I wish to

oh it seems possible yeah

for settings arrow, replace it with a keyboard with only numbers

Are you crazy? This will probably take more time to change and be more CPU-consuming than using the arrows

10:24 PM

huh only numbers

10:24 PM

like the Go to page thing?

tux-linux

like the Go to page thing?

exactly

https://stackoverflow.com/questions/59939409/how-to-expand-width-of-qspinbox-buttons oooh nice

How to expand width of QSpinBox buttons

I created a small GUI in QtCreator and had it converted to python to execute on a Raspberry Pi. issue is the QSpinBox buttons are only scaled by QSpinBox height. how can i increase the width of the

10:25 PM

It's actually possible

but inkbox is not written in python right?

10:25 PM

well its still qt

doesn't matter

10:26 PM

pyqt and qt share stylesheets

10:26 PM

and usually adapting the python code is not that hard

10:26 PM

Python and C++ are quite similar in some ways

for some things

how do i unload usb modules? they are busy

10:54 PM

resource busy

are you in usbnet?

10:56 PM

rmmod -f ?

tux-linux

are you in usbnet?

ssh

well, ssh with wifi?

kobo:~# rmmod -f configfs libcomposite u_ether usb_f_ecm usb_f_ecm_subset usb_f_rndis g_ether
rmmod: ERROR: could not remove 'configfs': Resource temporarily unavailable
rmmod: ERROR: could not remove module configfs: Resource temporarily unavailable
rmmod: ERROR: could not remove 'libcomposite': Resource temporarily unavailable
rmmod: ERROR: could not remove module libcomposite: Resource temporarily unavailable
rmmod: ERROR: could not remove 'u_ether': Resource temporarily unavailable
rmmod: ERROR: could not remove module u_ether: Resource temporarily unavailable
rmmod: ERROR: could not remove 'usb_f_ecm': No such file or directory
rmmod: ERROR: could not remove module usb_f_ecm: No such file or directory
rmmod: ERROR: could not remove 'usb_f_ecm_subset': No such file or directory
rmmod: ERROR: could not remove module usb_f_ecm_subset: No such file or directory
rmmod: ERROR: could not remove 'usb_f_rndis': Resource temporarily unavailable
rmmod: ERROR: could not remove module usb_f_rndis: Resource temporarily unavailable
rmmod: ERROR: could not remove 'g_ether': Resource busy
rmmod: ERROR: could not remove module g_ether: Resource busy
kobo:~#

10:56 PM

i already unloaded usbnet modules

10:56 PM

but those are still there

10:56 PM

im using wifi

i really don't know

10:57 PM

lsmod?

kobo:~# lsmod
Module                  Size  Used by
8189fs               1294344  0
sdio_wifi_pwr            593  0
g_ether                 5247  -1
usb_f_rndis            16183  1 g_ether
u_ether                12152  2 usb_f_rndis,g_ether
libcomposite           44430  2 usb_f_rndis,g_ether
configfs               23581  3 usb_f_rndis,libcomposite
kobo:~#

10:58 PM

some modules i could unloaded, those persist

kernel bug

-f doesn't work

if you want you can disable usbnet

10:58 PM

echo false > /boot/flags/USBNET_ENABLE

10:58 PM

and reboot

10:58 PM

it won't load the modules

yea those were usbnet modules

11:04 PM

they dissapered

1

yestarday i leaved it at 97% after removing usbnet, it was sleeping till now and i see 90%

4:49 PM

its acceptable + usbnet is eating battery like crazy

Good to know

4:50 PM

You don't have display debug flag enabled right?

you mean?

Look at wiki page

4:50 PM

Boot flags

4:51 PM

It prints serial output to screen rather than serial

disabled

Ok Because that eats a lot of battery

but the kernel has the quiet option removed, maybe this

4:51 PM

the wiki doesn't load for me

Which url?

4:54 PM

https://inkbox.ddns.net/wiki/index.php?title=Main_Page That doesn't work?

Main Page

oh, on ddg was this result: https://inkbox.ddns.net:40/wiki/index.php?title=Main_Page

tux-linux

https://inkbox.ddns.net/wiki/index.php?title=Main_Page That doesn't work?

this works

That's the new URL I changed some time ago I informed people in #chat

https://photos.app.goo.gl/nuwvQYKwfA1hQ1TL8 Also, a video I made a while ago

I want this boot animation (edited)

5:25 PM

does this consume energy all the time?

Yes

5:26 PM

I don't know why

5:27 PM

It takes 100% of the CPU when it's supposed to be idle for some reason

5:27 PM

But it's useful for debugging

sad

should echo "1" > /sys/power/state-extended disable the touch screen?

Szybet

yestarday i leaved it at 97% after removing usbnet, it was sleeping till now and i see 90%

But today it didn't dropped even one percentage

8:34 AM

I think there is a bug somewhere

8:35 AM

How do i gather all stdout / stderr from services to a file?

Maybe you can check /var/log/rc.log But I don't think it will be very helpful

I will need to edit the service files :\ (edited)

Szybet

should echo "1" > /sys/power/state-extended disable the touch screen?

And this?

That's to disable waking up the device with the touchscreen

1:29 PM

It only allows the power button/Hall sensor to wake it up

Oh okay

Now its confirmed When sleeping the touch screen is still on, and can awake the device

12:37 PM

Any fix for that? I think that /SYS/power/xxx should disable it

12:41 PM

I should create my own kernel

You need to echo 1 to power-extended

I think that the sleep service does that

1:20 PM

But its still is working

1:21 PM

The touch screen is very sensitive, even with a 1mm layer of PETG plastic

I don't know

Gutenberg hangs when Wi-Fi is off? On my devices it says Error as it should

1:48 PM

The reason I don't want to put a sync button is to limit the times/day one downloads the index from Gutenberg to prevent overloading the server

1:48 PM

The sync counter resets after one day

It tryies to sync and hangs

Maybe a button that only shows up after a day?

2:05 PM

I dont want accidentally sync the library

tux-linux

You need to echo 1 to power-extended

It didn't disabled the touch

12:33 PM

I did state-extended because power extended doesn't exist

yeah sorry my mistake

1:46 PM

@NiLuJe , do you know what's going on here? (touch waking up the device even if state-extended is set to 1)

I will check the kernel, maybe your config is missing something

maybe

1:50 PM

I uploaded the GUI bundle

tux-linux

you technically can but load busybox binary into ram before (just run busybox) without any arguments then busybox sync busybox reboot (edited)

kobo:/bin# busybox sync -ash: busybox: not found

5:21 PM

nope

Well idk

5:35 PM

Telnet into initrd

5:35 PM

Check INITRD_DEBUG boot flag

yes i know

5:53 PM

why powerpc etc

5:56 PM

and what is this: CONFIG_ATAGS_PROC

Well I didn't specifically enable these

Szybet

and what is this: CONFIG_ATAGS_PROC

Something for kexec to work (which I might do something with in the future)

before:

kobo:~# cat /sys/kernel/debug/suspend_stats 
success: 2
fail: 6
failed_freeze: 1
failed_prepare: 0
failed_suspend: 5
failed_suspend_late: 0
failed_suspend_noirq: 0
failed_resume: 0
failed_resume_early: 0
failed_resume_noirq: 0
failures:
  last_failed_dev:      228c000.epdc
                        228c000.epdc
  last_failed_errno:    -2
                        -2
  last_failed_step:     suspend
                        suspend

those fails are becouse i tryied to do something, but still hmm

6:38 PM

what is 228c000.epdc

6:40 PM

i2c to control the regulator for screen brightness and fbink

6:40 PM

interesting?

rebooted, now i get error -16 so "device and resource busy"

even with success the touchscreen wakes it up

i have disabled every wakeup file on the device:

kobo:/sys# find . -name "wakeup"
./devices/platform/soc/2000000.aips-bus/20cc000.snvs/20cc000.snvs:snvs-rtc-lp/power/wakeup
./devices/platform/soc/2000000.aips-bus/20c9000.usbphy/power/wakeup
./devices/platform/soc/2000000.aips-bus/2000000.spba-bus/2020000.serial/tty/ttymxc0/power/wakeup
./devices/platform/soc/2000000.aips-bus/20ca000.usbphy/power/wakeup
./devices/platform/soc/2100000.aips-bus/2184000.usb/power/wakeup
./devices/platform/soc/2100000.aips-bus/2184000.usb/ci_hdrc.0/power/wakeup
./devices/platform/soc/2100000.aips-bus/2194000.usdhc/power/wakeup
./devices/platform/soc/2100000.aips-bus/2190000.usdhc/power/wakeup
./devices/platform/soc/2100000.aips-bus/21ec000.serial/tty/ttymxc2/power/wakeup
./devices/platform/soc/2100000.aips-bus/21f8000.i2c/i2c-3/3-0032/rtc_ricoh619/power/wakeup
./devices/platform/soc/2100000.aips-bus/21f8000.i2c/i2c-3/3-0032/ricoh619-battery/power_supply/acpwr/power/wakeup
./devices/platform/soc/2100000.aips-bus/21f8000.i2c/i2c-3/3-0032/ricoh619-battery/power_supply/mc13892_charger/power/wakeup
./devices/platform/soc/2100000.aips-bus/21f8000.i2c/i2c-3/3-0032/ricoh619-battery/power_supply/mc13892_bat/power/wakeup
./devices/platform/soc/2100000.aips-bus/21f8000.i2c/i2c-3/3-0032/ricoh619-battery/power_supply/usbpwr/power/wakeup
./devices/platform/gpio-keys/power/wakeup

its still waking up

i have no idea, here is some documentation that isin't working: https://www.kernel.org/doc/html/v4.14/driver-api/pm/devices.html

this is disabled:

8:35 PM

:(

why i cant boot now any kernel

message.txt

22.37 KB

12:13 AM

here is nickel kernel: Starting kernel ...

/cpus/cpu@0 missing clock-frequency property
sy7636-pmic sy7636-pmic: request vcom gpio failed (-16)!
sy7636-pmic sy7636-pmic: request powerup gpio failed (-16)!
syscon-poweroff 20cc000.snvs:snvs-poweroff: pm_power_off already claimed 80021934 ntx_machine_poweroff
PMU: ricoh61x_battery_probe : version is RICOH61x_BATTERY_VERSION: 2014.02.21 V3.1.0.0-Solution1 2015/02/09
==== calc_soc_by_voltageMethod E60U22 ====
imx6ul-pinctrl 2290000.iomuxc-snvs: pin MX6ULL_PAD_SNVS_TAMPER3 already requested by 21f8000.i2c; cannot claim for gpio-keys
imx6ul-pinctrl 2290000.iomuxc-snvs: pin-5 (gpio-keys) status -22
imx6ul-pinctrl 2290000.iomuxc-snvs: could not request pin 5 (MX6ULL_PAD_SNVS_TAMPER3) from group gpio_snvs_keys_grp  on device 22900s
gpio-keys gpio-keys: Error applying setting, reverse things back
Kernel panic - not syncing: No working init found.  Try passing init= option to kernel. See Linux Documentation/init.txt for guidanc.
---[ end Kernel panic - not syncing: No working init found.  Try passing init= option to kernel. See Linux Documentation/init.txt fo.
imx_epdc_v2_fb 228c000.epdc: Cleared LUT complete for init update

nickel image works god i was scared that i damaged the cpu

12:57 AM

now i have a important question

12:57 AM

what is /dev/sdcard in inkbox?

12:57 AM

@tux-linux

no idea?

Szybet

why i cant boot now any kernel

sudo env GITDIR="${PWD}" scripts/make_devicenodes.sh

5:29 PM

this will solve your problem

100%?

Well yeah?

i will test it

no, not really

12:12 AM

its the same

Did you recompile?

yes

Boot log?

message.txt

23.34 KB

12:21 AM

this is the default config kernel

12:25 AM

something is compiling wrong in my opinion

kernel build log:

message.txt

221.64 KB

12:42 AM

u-boot is saying something interesting: or maybe not

eBR-1A # boot
switch to partitions #0, OK
mmc0 is current device
mmc read 0x80800000 0x14000 0x5000

MMC read: dev # 0, block # 81920, count 20480 ... 20480 blocks read: OK
Booting from mmc ...
mmc read 0x83000000 0x505 0x1

MMC read: dev # 0, block # 1285, count 1 ... 1 blocks read: OK
dtb size = 41589@83000000
mmc read 0x83000000 0x506 0x57

MMC read: dev # 0, block # 1286, count 87 ... 87 blocks read: OK

 hwcfgp=8ffffe00,pcb=81,customer=9

ntx_gpio_get_value(404) : error parameter ! null ptr !
ESDin=0,UPGKey=-1,PWRKey=0,USBin=0x1,BootESD=0,MenuKey=0
mmc read 0x8ffffc00 0x37ff 0x1

MMC read: dev # 0, block # 14335, count 1 ... 1 blocks read: OK
mmc read 0x8ff71a00 0x3800 0x472

MMC read: dev # 0, block # 14336, count 1138 ... 1138 blocks read: OK
mmc read 0x8ff71800 0x405 0x1

MMC read: dev # 0, block # 1029, count 1 ... 1 blocks read: OK
[WARNING] Binaries load sequence should Lo->Hi !
mmc read 0x8ff71400 0x406 0x3

MMC read: dev # 0, block # 1030, count 3 ... 3 blocks read: OK
Kernel RAM visiable size=255M->255M
ntx_gpio_init():R [ERROR] GPIO group number error (0)!!
ntx_gpio_init():B [ERROR] GPIO group number error (0)!!
hwcfg rootfstype : 2
hwcfg partition type : 2,bootmode=0
ntx_prebootm : cmd=setenv bootargs ${bootargs}  hwcfg_p=0x8ffffe00 hwcfg_sz=110 waveform_p=0x8ff71a00 waveform_sz=582529 ntxfw_p=0x8t
Kernel image @ 0x80800000 [ 0x000000 - 0x9b88d0 ]
## Flattened Device Tree blob at 83000000
   Booting using the fdt blob at 0x83000000
   Using Device Tree in place at 83000000, end 8300d274

Starting kernel ...

(edited)

i have no idea

it cant go to the point of Linux inkbox 4.1.15 #1 SMP PREEMPT Sat Feb 19 21:21:19 CET 2022 armv7l GNU/Linux

1:11 AM

idk, any suggestions @tux-linux

i have tryied removing some files from initrd/n306/opt/ becouse the previous problems were avoided in this way, but nothing

1:36 AM

and the kernel binary is still 9,8 Mb after all

1:36 AM

Can you try completely deleting the repo and cloning it again?

tux-linux

sudo env GITDIR="${PWD}" scripts/make_devicenodes.sh

Don't forget to do this again before you compile it

tux-linux

Can you try completely deleting the repo and cloning it again?

I did this already, and nothing

ls initrd/n306/dev

The 3 nodes are there, i checked it

cat kernel/config/config-n306 | grep CMD (edited)

tux-linux

ls initrd/n306/dev

% ls -l initrd/n306/dev
razem 0
crw-r--r-- 1 root root   5,  1 03-26 11:51 console
crw-r--r-- 1 root root   1,  3 03-26 11:51 null
crw-r--r-- 1 root root 207, 16 03-26 11:51 ttymxc

tux-linux

cat kernel/config/config-n306 | grep CMD (edited)

i recloned the repo...

% cat kernel/config/config-n306| grep CMD
# CONFIG_BLK_CMDLINE_PARSER is not set
CONFIG_CMDLINE="console=ttymxc0,115200 rootwait rw no_console_suspend hwcfg_p=0x8ffffe00 hwcfg_sz=110 waveform_p=0x8ff71a00 waveform_sz=582529 ntxfw_p=0x8ff71400 ntxfw_sz=1034 mem=255M boot_port=1 rootfstype=ramfs root=/dev/ram0 rdinit=/sbin/init printk.time quiet"
# CONFIG_CMDLINE_FROM_BOOTLOADER is not set
# CONFIG_CMDLINE_EXTEND is not set
CONFIG_CMDLINE_FORCE=y
# CONFIG_BCMDHD is not set
CONFIG_FB_CMDLINE=y
11:52 szybet@ArchSzybet /mnt/HDD/Project/Public/kobo-nia-audio/inkbox/kernel
%

Szybet

% ls -l initrd/n306/dev
razem 0
crw-r--r-- 1 root root   5,  1 03-26 11:51 console
crw-r--r-- 1 root root   1,  3 03-26 11:51 null
crw-r--r-- 1 root root 207, 16 03-26 11:51 ttymxc

There is ttymxc0, right?

11:53 AM

Not just ttymxc

the command to compile:

env GITDIR="${PWD}" TOOLCHAINDIR="${PWD}/toolchain/arm-nickel-linux-gnueabihf/" THREADS=$(($(nproc)*2)) TARGET=arm-nickel-linux-gnueabihf scripts/build_kernel.sh n306 root

the command to write: sudo dd if=zImage-root of=/dev/sdd bs=512 seek=81920

tux-linux

There is ttymxc0, right?

yes

Did you change the location of the initrd folder in the kernel config? Is /home/build/inkbox a working symlink?

yes, otherwise it doesn't compile at all

tux-linux

Did you change the location of the initrd folder in the kernel config? Is /home/build/inkbox a working symlink?

initrd? where to change it?

11:57 AM

its in the kernel repo

Szybet

initrd? where to change it?

It's fine Since you have a symlink, it works

the kernel size is always 9,8 M

is the dd kernel command good?

12:22 PM

i first did dd if=inkbox_maybe_working.img of=kernel_inkbox_working count=10000000 skip=81920 iflag=skip_bytes,count_bytes to get the working kernel from a working inkbox image

12:23 PM

then sudo dd if=kernel_inkbox_working of=/dev/sdd bs=512 seek=81920

12:23 PM

but nothing

12:23 PM

no change

Maybe you could compile the kernel... there is maybe something wrong with my linux on pc

its 83000000 booting from but we are writing to 81920

Why root at the end od the command and not std, it is like on your github

83000000 is the memory address

2:35 PM

Yes, I can try

tux-linux

83000000 is the memory address

So?

what the fuck is going on

3:14 PM

i compiled the kernel on my brothers laptop

3:15 PM

same device, same os but its a bit older and with less packages installed

3:15 PM

and the kernel booted

1

3:15 PM

its time to use docker for kernel build

3:16 PM

Szybet

its time to use docker for kernel build

Maybe lxc could be better

maybe

3:50 PM

but still, i want my wasted time back

How much time since you rebooted your laptop

Szybet

but still, i want my wasted time back

I had no way of knowing this

tux-linux

I had no way of knowing this

i know i know, its just computer problems

tux-linux

How much time since you rebooted your laptop

i have tryied yestarday, rebooted today, rebooted, and again (edited)

3:54 PM

so no, no reboot has to do with it

I think I will start making the update packages/images for 1.9 now, unless there's something absolutely critical to solve?...

1

As for the sleep/touchscreen issues, I don't know what to do, even KOReader seems to do it all the time: https://github.com/koreader/koreader/blob/fb45576429b4302cd7dd7f09999206a5d60e9960/frontend/device/kobo/device.lua#L762-L775

i think i can figure it out adding one kernel option with the touchscreen

Szybet

this is disabled:

this?

yes

5:37 PM

then I echo disable > wakeup to the touchscreen device

it works? (edited)

Szybet

Why root at the end od the command and not std, it is like on your github

https://inkbox.ddns.net/wiki/index.php?title=General_information#Standard_and_Rooted Default is std

General information

oh thats it

Szybet

oh thats it

what?

std and root option

oh ok

Szybet

i think i can figure it out adding one kernel option with the touchscreen

can you try it?

i will, first i tinker with lxc to get the kernel to even boot

hmm

5:55 PM

do you need a kernel? i can compile one, just give me the config file

i will compile it on my own

1

6:07 PM

lxc is horrible

then maybe try a vm?

documentation is for old commands, outdated

6:07 PM

forums too

6:07 PM

just npo

i will add this to the guide later: install distrobox: https://github.com/89luca89/distrobox create a image:

distrobox-create --image debian:latest
distrobox-enter debian-latest

install some packages:

 
sudo apt update
sudo apt upgrade
sudo apt install bc git build-essential u-boot-tools libncurses5-dev libncursesw5-dev

create the symlink:

sudo mkdir -p /home/build/inkbox/
sudo ln -s path_to_kernel_repo /home/build/inkbox/kernel

thats it, now cpp magic won't waste your time (edited)

So it boots fine?

now, yes

Ok nice

look:

kobo:~# cat /sys/kernel/debug/wakeup_sources
name            active_count    event_count     wakeup_count    expire_count    active_since    total_time      max_time        last_change     prevent_suspend_time
gpio-keys       0               0               0               0               0               0               0               5957            0
mc13892_charger 1               1               0               0               0               8               8               5701            0
usbpwr          1               1               0               0               0               7               7               5700            0
acpwr           1               1               0               0               0               5               5               5698            0
mc13892_bat     14              14              0               0               0               500             188             561099          0
20cc000.snvs:snvs-rtc-lp        0               0               0               0               0               0               0               5603       0
alarmtimer      0               0               0               0               0               0               0               4996            0
rtc_ricoh619    0

i disabled them all:

kobo:~# cat /sys/kernel/debug/wakeup_sources
name            active_count    event_count     wakeup_count    expire_count    active_since    total_time      max_time        last_change     prevent_suspend_time
alarmtimer      0               0               0               0               0               0               0               4996

but its still waking up

7:39 PM

well, this did nothing

strange

7:40 PM

kobo kernels are, by definition, strange and unusual in all sorts of not-so-nice ways

7:40 PM

I don't know what to tell you

there are some new files:

7:41 PM

kobo:~# ls /sys/power/
pm_async           pm_print_times     state              wake_lock          wakeup_count
pm_freeze_timeout  pm_test            state-extended     wake_unlock
kobo:~# `

7:42 PM

but i cant find documentation for wake* files

7:45 PM

i checked on nickel, it does not happen

7:45 PM

wakeup becouse of touchscreen

maybe it puts it to sleep automatically if it somehow detects that the wakeup source was the touchscreen?

7:46 PM

but that's soo dirty

i think kobo developers are not dumb

but, sometimes, they seem to be in a big hurry

well

7:48 PM

what manages sleep on nickel, i would want to know

Is there a way to know which device has awaken the device?

i don't know a lot about the wakeup system honestly

maybe udev has something with it?

8:24 PM

and this, on a kobo ereader:

8:24 PM

nice

where do you see that? (edited)

nickel filesystem

8:26 PM

i have a image

which file

Szybet

maybe udev has something with it?

probably

tux-linux

which file

a udev rule

1

8:32 PM

is there a way to disable the touchscreen?... anything...

1

GPIO's seem to be "written in hurry" too

8:44 PM

gpio_keys dont show up in ./bus/platform/devices/gpio-keys

my last idea: unbind the touchscreen driver

message.txt

2.48 KB

8:54 PM

but which device is it ._ .

8:54 PM

ntx_event0 not

have a look at the kernel source maybe? (edited)

how do i know which files are actually used while building?

8:55 PM

if i could seperate them, then maybe

lemme find a pcb picture

????????????...

to find the touchscreen microcontroller

for the driver unbind? i think it will not help

it will help to find the touchscreen driver name, probably

tux-linux

lemme find a pcb picture

do you have one?

there are like 5 main chips on the device, and 3 small, one smaller than a diode

tux-linux

do you have one?

i have the kobo physically

8:58 PM

but yea, wait

1

Szybet

there are like 5 main chips on the device, and 3 small, one smaller than a diode

only on one side*

8:58 PM

https://github.com/Szybet/kobo-nia-audio/tree/main/images/nia

kobo-nia-audio/images/nia at main · Szybet/kobo-nia-audio

Kobo nia ereader with audio added. Contribute to Szybet/kobo-nia-audio development by creating an account on GitHub.

9:01 PM

Szybet

Click to see attachment 🖼️

okay discord, cool

9:01 PM

cool...

9:02 PM

now

9:02 PM

those are all IC on the device

9:03 PM

oh, and the big one on ribbon up right

try to see if some controller name matches something in /sys/devices/platform, here's it's too small to read

Kobo_nia.pdf

7.55 MB

9:31 PM

this won't help much

9:31 PM

but its pretty cool

9:31 PM

why is there a rainbow in those codes

9:32 PM

Kobo_nia.pdf

7.55 MB

9:32 PM

here is a better one

FOUND IT

9:44 PM

pure luck

9:44 PM

/sys/bus/i2c/drivers/elan-touch

9:45 PM

its /sys/devices/platform/soc/2100000.aips-bus/21a0000.i2c/i2c-0/0-0015

9:45 PM

kobo:/sys/bus/i2c/drivers/elan-touch# echo 0-0015 > unbind 
kobo:/sys/bus/i2c/drivers/elan-touch# echo 0-0015 > bind
ash: write error: No such device

this just cant be easy :C

9:46 PM

and the kernel panicked or smth

9:46 PM

message.txt

9.36 KB

is the ereader still working? (edited)

it rebooted and yes

10:12 PM

why do you ask :)?

no, I wanted to know if the kernel 'panic' had an effect on the working system

10:13 PM

like, if it went by without issue or it required a reboot

it forced a reboot

hmm ok that is critical

i see a file called delete_device in the i2c parrent, i think this is the proper way to remove a device

kobo:/sys/devices/platform/soc/2100000.aips-bus/21a0000.i2c/i2c-0# ls
0-0015         0-0062         delete_device  name           of_node        subsystem
0-0036         0-0068         i2c-dev        new_device     power          uevent

but how to use it

10:14 PM

where documentation

what makes you think that there is documentation available ? ;p

Szybet

i see a file called delete_device in the i2c parrent, i think this is the proper way to remove a device

kobo:/sys/devices/platform/soc/2100000.aips-bus/21a0000.i2c/i2c-0# ls
0-0015         0-0062         delete_device  name           of_node        subsystem
0-0036         0-0068         i2c-dev        new_device     power          uevent

but how to use it

try it?

its the linux kernel... its opensource... :(

tux-linux

try it?

invalid argument

what did you write?

0-0015 or elan-touch

10:15 PM

maybe the hex address of the device

10:15 PM

but i need to find it

https://stackoverflow.com/questions/58843718/how-can-i-dynamically-remove-or-retrigger-the-probing-an-eeprom-via-sysfs

How can I dynamically remove or retrigger the probing an EEPROM via...

I am wondering what the proper way is to remove EEPROMs dynamically. An alternate approach that will also work is telling the device to re-do the probe. I tried just echoing it to delete_device, ...

i was there already ;p

maybe just 0x15 ?

the first option doesn't work as you see

tux-linux

maybe just 0x15 ?

tryied it, nope

10:16 PM

but here is another error:

kobo:/sys/devices/platform/soc/2100000.aips-bus/21a0000.i2c/i2c-0# echo 0x15 > delete_device 
ash: write error: No such file or directory

https://erlerobotics.gitbooks.io/erle-robotics-erle-brain-a-linux-brain-for-drones/content/en/tutorials/i2c.html

How to instantiate I2C from the userspace · Erle Robotics: Erle-bra...

10:17 PM

hmm

tux-linux

https://erlerobotics.gitbooks.io/erle-robotics-erle-brain-a-linux-brain-for-drones/content/en/tutorials/i2c.html

i was there... wait there is the delete too

10:17 PM

hm, good now what is the proper address

look in the dts in the kernel source

10:19 PM

there you will probably find it

and dts is?

DeviceTree Source

10:21 PM

what's the identifier written on the PCB?

10:21 PM

ok got it

is it imporant?

do you have ddr3 ram?

i think its 2 yes: https://github.com/Szybet/kobo-nia-audio/blob/6503a0fa8c29f80764087a916ef1f8bb350310a8/logs/ntx_hwconfig#L32

ok so take arch/arm/boot/dts/imx6ull-ntx.dtsi

thats a cool file

but it seems it isn't referenced maybe arch/arm/boot/dts/imx6ull-ntx-ddr3.dts (edited)

10:25 PM

though it still seems to be 0x15

yes

10:26 PM

maybe removing them is disabled? lets look to kernel config

10:28 PM

here is nothing

10:28 PM

tux-linux

though it still seems to be 0x15

wait... what are those gpio's there?

10:28 PM

i2c pins?

Maybe

10:34 PM

Be careful with that, sometimes you can physically damage the device if you do the wrong thing

yes but actually yes and no... most important if the regulators have good maximum voltages assing (edited)

hmm

10:42 PM

HMMMMM

10:43 PM

modules

10:43 PM

10:44 PM

well, first you need to give instructions how to create modules.sqash

10:44 PM

just modules_install and then the whole directory into squash?

No, why?

?

Szybet

Click to see attachment 🖼️

That will probably conflict with the current driver

i knwo, but still its something

Szybet

Click to see attachment 🖼️

at default this is built in

10:49 PM

i want to module it

10:49 PM

so how do i make modules.squash

Before, just try to insmod it without putting it into modules.sqsh

um, so MODULES_INSTALL, copy the files to device?

Yeah

11:14 PM

SCP, etc

well, tommorow

1

Szybet

FOUND IT

Can you see if it panics again now, or was it just a coincidence?

it panics multiple times

Ok

actually, if i try to access the file system then it panicks

Hopefully you'll find something that will make it work

Szybet

actually, if i try to access the file system then it panicks

Well, doesn't really help here lol

ls: can't open '/': Connection aborted

1

11:17 PM

and after this there is the kernel panik

something something wakeup pins in the dts maybe?

1:16 AM

No experience with the elan driver, I think it's the one found on the Sage & Elipsa, though, and I'm pretty sure it's PM'ed properly on stock images there.

1

  Building modules, stage 2.
  MODPOST 28 modules
ERROR: "gSleep_Mode_Suspend" [drivers/input/touchscreen/elan_touch_i2c.ko] undefined!
ERROR: "gIsCustomerUi" [drivers/input/touchscreen/elan_touch_i2c.ko] undefined!
ERROR: "gptHWCFG" [drivers/input/touchscreen/elan_touch_i2c.ko] undefined!
make[1]: *** [scripts/Makefile.modpost:90: __modpost] Error 1
make: *** [Makefile:1103: modules] Error 2
[szybet@debian-latest linux-4.1.15-n306]$

11:37 AM

:/

11:41 AM

this file does not exist at all

#include "../../../arch/arm/mach-imx/ntx_hwconfig.h"
extern volatile NTX_HWCONFIG *gptHWCFG;

weird

11:55 AM

this can be resolved, but i dont know cpp to good for this

11:55 AM

this is pure c maybe

11:58 AM

so 3 problems: 3 variables are undefined in kernel/kernel/linux-4.1.15-n306/drivers/input/touchscreen/elan_touch_i2c.c: line 768: extern int gSleep_Mode_Suspend; line 50 extern int gIsCustomerUi; line 27 ( this one is the weirdiest )

#include "../../../arch/arm/mach-imx/ntx_hwconfig.h"
extern volatile NTX_HWCONFIG *gptHWCFG;

here is the full file

elan_touch_i2c.c

39.14 KB

disabling the touchscreen would be easy with udev... https://ask.fedoraproject.org/t/disabling-device-with-udev/1975

Szybet

disabling the touchscreen would be easy with udev... https://ask.fedoraproject.org/t/disabling-device-with-udev/1975

Try, udev is supposedly there in the rootfs

yes, and it doesn't work

What specifically?

udevadm test --action="remove" /devices/platform/ntx_event0/input/input

2:11 PM

well, udev isin't mean't for this

2:12 PM

we need to try with the modules

2:12 PM

or

2:12 PM

in the files elan* are many thing related to wakeup, i dont really understand them

Neither do I And neither do I have an idea about what in the world is happening with undefined variables. NTX/Kobo kernel code is a mess.

2:15 PM

Maybe you can try with gpio pins?

you mean?

2:15 PM

the touch screen is i2c

interesting bug

[  250.760770] Freezing of tasks failed after 20.008 seconds (1 tasks refusing to freeze, wq_busy=0):
[  250.769848] udevd           D 8056fa00     0   814      1 0x00000001
[  250.769897] [<8056fa00>] (__schedule) from [<8056fd8c>] (schedule+0x40/0x98)
[  250.769917] [<8056fd8c>] (schedule) from [<801bcfdc>] (__fuse_request_send+0x13c/0x2c8)
[  250.769938] [<801bcfdc>] (__fuse_request_send) from [<801bfe08>] (fuse_simple_request+0xc8/0x1dc)
[  250.769958] [<801bfe08>] (fuse_simple_request) from [<801c2208>] (fuse_lookup_name+0xb8/0x16c)
[  250.769975] [<801c2208>] (fuse_lookup_name) from [<801c22f4>] (fuse_lookup+0x38/0xdc)
[  250.769995] [<801c22f4>] (fuse_lookup) from [<800ef928>] (lookup_real+0x20/0x4c)
[  250.770015] [<800ef928>] (lookup_real) from [<800f082c>] (__lookup_hash+0x34/0x3c)
[  250.770033] [<800f082c>] (__lookup_hash) from [<800f086c>] (lookup_slow+0x38/0xa4)
[  250.770050] [<800f086c>] (lookup_slow) from [<800f42e0>] (path_lookupat+0x5b8/0x704)
[  250.770066] [<800f42e0>] (path_lookupat) from [<800f4448>] (filename_lookup+0x1c/0x64)
[  250.770080] [<800f4448>] (filename_lookup) from [<800f5840>] (user_path_at_empty+0x54/0x7c)
[  250.770096] [<800f5840>] (user_path_at_empty) from [<800f587c>] (user_path_at+0x14/0x1c)
[  250.770112] [<800f587c>] (user_path_at) from [<800ebb08>] (vfs_fstatat+0x44/0x98)
[  250.770131] [<800ebb08>] (vfs_fstatat) from [<800ec1e0>] (SyS_stat64+0x14/0x30)
[  250.770151] [<800ec1e0>] (SyS_stat64) from [<8000f440>] (ret_fast_syscall+0x0/0x3c)
[  250.770181] 
[  250.770190] Restarting tasks ... done.

and: kobo:~# [ 250.760770] Freezing of tasks failed after 20.008 seconds (1 tasks refusing to freeze, wq_busy=0):

3:23 PM

I did this

3:23 PM

i dont know how

3:24 PM

._ .

3:25 PM

it works

3:25 PM

bruhhhhh

3:25 PM

i will make a pull request later

What did you do??

Random // in elan drivers where wake enabled was

6:52 PM

I got to this becouse od proc interrupts and in the irq in files

lool

make a pull request when you can

maybe not today, tommorow

Can you pull the kernel repo and see if the kernel still boots with the new 203K binary (qr-to-png) I added?

tommorow propably

if i compile it for you?

also the kernel size... max is 10,5M currently its 9,8M, and i want to add alsa and soundcard support

8:51 PM

this doesn't look good for the future

tux-linux

if i compile it for you?

then propably today

Szybet

this doesn't look good for the future

0.7M is probably already too small

8:52 PM

Maybe you can put them in a [p1]/modules folder, I don't know

what is next to the kernel?

partition 3 iirc

waveform?

tux-linux

partition 3 iirc

not good

yeah, well, when I set up a partition layout I didn't think the kernels would become so huge

Szybet

then propably today

compiling right now

there are 2 options: 1 we make the partition layout in a other way now 2 in the future we will do this anyway, e-readers are becoming more complicated and have more features. Even never kernels are bigger

the thing is that I don't want to re-upload a system image for existing supported models

now not many people use inkbox, it will not hurt many. in the future it will

i guess 10.5M is also really huge for a kernel on an embedded device

8:56 PM

when you think about it, it eats 10M of your RAM

8:57 PM

so it is probably not a good idea to keep it growing

8:57 PM

we'll have to find other ways to do what we want to do

modules etc on a partition

the thing is that each device has different modules

8:58 PM

painful to update

8:58 PM

but maybe

just update with the rootfs

8:58 PM

+ this can be done with a script

i'll see

9:00 PM

https://drive.google.com/file/d/1cUnJFokT8qt3CcANmFAqE62FaLb2-mbh/view?usp=sharing

zImage-root

did you get any success?

it boots

ok, nice

9:20 PM

thanks

is this normal that when going to sleep, when the 10-20s process is going if i click the touchscreen it cancels the sleeping? (edited)

No

12:23 AM

But maybe it is good

I updated the GUI bundle with latest features, including 2048 and QReversi games http://23.163.0.39/bundles/private/update-n306.isa

6:33 AM

I modified 2048, now it is more optimized for eReaders https://github.com/Kobo-InkBox/2048

6:34 AM

tux-linux

But maybe it is good

In what way

i don't know (edited)

tux-linux

I updated the GUI bundle with latest features, including 2048 and QReversi games http://23.163.0.39/bundles/private/update-n306.isa

I will test it today

1

tux-linux

i don't know (edited)

Then not good :p

well, you fix it, i've had enough nightmares with sleep issues (yeah, get it? o_O) ;p (edited)

2

tux-linux

I modified 2048, now it is more optimized for eReaders https://github.com/Kobo-InkBox/2048

and I must have played at least half an hour total on it since it is implemented this game is really addictive lol

tux-linux

well, you fix it, i've had enough nightmares with sleep issues (yeah, get it? o_O) ;p (edited)

I will redesign it completly

tux-linux

and I must have played at least half an hour total on it since it is implemented this game is really addictive lol

yes, it definitely is lol

./wake_standby.sh: line 14: echo: write error: Operation not permitted
[FBInk] Couldn't find a Kobo version tag (onboard unmounted or not running on a Kobo?)!
[FBInk] Detected a Kobo Nia (382 => Luna @ Mark 7)
[FBInk] Enabled Kobo Mark 7 quirks
[FBInk] Clock tick frequency appears to be 100 Hz
[FBInk] Screen density set to 212 dpi
[FBInk] Variable fb info: 758x1024, 16bpp @ rotation: 3 (Counter Clockwise, 270°)
[FBInk] Fixed fb info: ID is "mxc_epdc_fb", length of fb mem: 3145728 bytes & line length: 1536 bytes
[FBInk] Canonical rotation: 0 (Upright, 0°)
[FBInk] Fontsize set to 16x16 (IBM base glyph size: 8x8)
[FBInk] Line length: 47 cols, Page size: 64 rows
[FBInk] Pen colors set to #000000 for the foreground and #FFFFFF for the background
[FBInk] Couldn't find a Kobo version tag (onboard unmounted or not running on a Kobo?)!
[FBInk] Detected a Kobo Nia (382 => Luna @ Mark 7)
[FBInk] Enabled Kobo Mark 7 quirks
[FBInk] Clock tick frequency appears to be 100 Hz
[FBInk] Screen density set to 212 dpi
[FBInk] Variable fb info: 758x1024, 16bpp @ rotation: 3 (Counter Clockwise, 270°)
[FBInk] Fixed fb info: ID is "mxc_epdc_fb", length of fb mem: 3145728 bytes & line length: 1536 bytes
[FBInk] Canonical rotation: 0 (Upright, 0°)
[FBInk] Fontsize set to 16x16 (IBM base glyph size: 8x8)
[FBInk] Line length: 47 cols, Page size: 64 rows
[FBInk] Pen colors set to #000000 for the foreground and #FFFFFF for the background
Displaying image '/tmp/dump.png' @ column 0 + 0px, row 0 + 0px (scaling: 0x0, H align: 0, V align: 0, inverted: N, flattened: N, waveform: AUTO, HW dithering: PASSTHROUGH, SW dithered: N, nightmode: N, skip refresh: N)

and at line 14: echo "mem" > /sys/power/state

9:05 PM

Interesting

9:07 PM

propably the FBINK is related to the script later

9:07 PM

but still

[ 2341.748944] Freezing user space processes ... (elapsed 0.001 seconds) done.
[ 2341.750868] Freezing remaining freezable tasks ... (elapsed 0.001 seconds) done.
[ 2341.752192] PM: Entering mem sleep
[ 2341.755135] [evdev_check_event-1232] event1 head D4, tail 3A
[ 2341.755169] dpm_run_callback(): input_dev_suspend+0x0/0xac returns -1
[ 2341.761767] PM: Device input1 failed to suspend: error -1
[ 2341.767180] PM: Some devices failed to suspend, or early wake event detected
[ 2341.847891] PM: resume of devices complete after 73.565 msecs
[ 2341.849635] PM: resume devices took 0.080 seconds
[ 2341.849653] PM: Finishing wakeup

9:17 PM

this touch device...

On other devices it seems to be the framebuffer

here its only the touch

10:30 PM

well, i can get the error message and if it appears it will try one more time to go to sleep

10:31 PM

how do i add services to launch at boot, and commit it? where is the openrc config fle located

/etc/runlevels/default

10:40 PM

Symlink your service there

okay

10:40 PM

i will redesign the sleep thing

I discovered other issues with the reader; it seems that it still doesn't work when I open a PDF then an ePUB

Szybet

i will redesign the sleep thing

Good luck

tux-linux

I discovered other issues with the reader; it seems that it still doesn't work when I open a PDF then an ePUB

i said that before

10:41 PM

i had issues with epubs only but yes

Yeah, and I fixed it

10:41 PM

Strange

tux-linux

I discovered other issues with the reader; it seems that it still doesn't work when I open a PDF then an ePUB

Also I have strange scaling in PDFs when I click on next page button Will try to investigate this tonight

tux-linux

Yeah, and I fixed it

Right?

tux-linux

Also I have strange scaling in PDFs when I click on next page button Will try to investigate this tonight

i had scalling issues in epubs too. no images, not enough text and the "end of document" toast appeared after every page change. i think it happened in the newest gui bundle

tux-linux

Right?

yes, but im saying now about another bug

Szybet

i had scalling issues in epubs too. no images, not enough text and the "end of document" toast appeared after every page change. i think it happened in the newest gui bundle

Uh, not the end of document thing. I read on InkBox today and ePUB is fine

10:47 PM

Not enough text, you can change ePUB page size in Settings

tux-linux

Not enough text, you can change ePUB page size in Settings

i tryied, nothing

And images are simply not supported, sorry. It's a limitation of the viewing engine

Szybet

i tryied, nothing

Really?

yes

For me it works

tux-linux

And images are simply not supported, sorry. It's a limitation of the viewing engine

??? mudpdf is used in koreader and plato and i think its working there

It's not used in the same way

10:51 PM

The GUI uses qtextedit

10:51 PM

It displays a generated html by mupdf

qtextedit image another qtextedit

10:52 PM

this could be done?

10:53 PM

i think mupdf gives a reference to the image

10:53 PM

if its html, maybe just use qtbrowser or something like that?

No, qtwebengine doesn't work on kobos

10:54 PM

Also even if it did it would consume way too much battery

Szybet

qtextedit image another qtextedit

then this?

Szybet

this could be done?

Maybe

i will use something like this in sanki

1

can i use fbink command to write many times one thing? will it damage the e-ink?

11:06 PM

also about e-inks. do magnets have a effect on them?

11:07 PM

what does this line do

11:08 PM

chroot /kobo /usr/bin/fbgrab "/external_root/tmp/dump.png" this command does nothing?...

Takes a screenshot

Szybet

Click to see attachment 🖼️

Empties file

Szybet

can i use fbink command to write many times one thing? will it damage the e-ink?

I haven't seen damage, but apparently it is possible

Szybet

also about e-inks. do magnets have a effect on them?

From what I've read, strong ones do have some effect

tux-linux

Takes a screenshot

this makes sense

tux-linux

From what I've read, strong ones do have some effect

how strong ones?

11:33 PM

i want to make a modular etui

Szybet

how strong ones?

Don't know The magnet in cases is ok enough for sure

/opt/bin/fbink/fbink -k -f -q
/opt/bin/fbink/fbink -t regular=/etc/init.d/splash.d/fonts/resources/inter-b.ttf,size=20 "Sleeping" -m -M -q

can i put this just before going to sleep?

Please use this script syntax: "${VAR}" for var expansion (double quote unless you know it will be an int) "$(command)" for invoking command in a var

Szybet

/opt/bin/fbink/fbink -k -f -q
/opt/bin/fbink/fbink -t regular=/etc/init.d/splash.d/fonts/resources/inter-b.ttf,size=20 "Sleeping" -m -M -q

can i put this just before going to sleep?

I think? But not after sleep 10

tux-linux

Please use this script syntax: "${VAR}" for var expansion (double quote unless you know it will be an int) "$(command)" for invoking command in a var

sure

1

https://github.com/Kobo-InkBox/rootfs/blob/master/etc/init.d/update.sh This is a good example of the syntax I want Everything is there, except some places when I forgot to double quote

rootfs/update.sh at master · Kobo-InkBox/rootfs

InkBox OS root filesystem. Contribute to Kobo-InkBox/rootfs development by creating an account on GitHub.

okay

11:40 PM

11:41 PM

where does this else happen?

Yes it is needed

11:41 PM

When there is an event in event0 but it is not the power button or sleep cover

11:41 PM

Sometimes it can be the gyroscope for example

tux-linux

Yes it is needed

Otherwise you will get a memory leak at some point

but after detecting something it will continue to run the code after it?

It continues/restarts the while loop (edited)

11:42 PM

You know what inotifywait does, right?

Wait for a particular event on a file or set of files.

Yes So it waits for an input event then executes the code

so only for KEY_POWER and F1, okay

Yes

can i write to /tmp/ some data, sync and sleep 2 and then mem > state?

well, one thing and i think it will be perfect

2:25 AM

how do i specify

fbink -q -p -m -y 30 -S 3 -B WHITE -t regular=/etc/init.d/splash.d/fonts/resources/inter-b.ttf,size=10 "don't touch the screen"

the place where to spawn the text using those fonts? i tryied with size=10,bottom=10 etc but nothing worked

2:27 AM

ugh, brightness doesn't work

2:30 AM

well, tommorow

2:31 AM

but its now reacting instant for the power button

2:31 AM

Szybet

how do i specify

fbink -q -p -m -y 30 -S 3 -B WHITE -t regular=/etc/init.d/splash.d/fonts/resources/inter-b.ttf,size=10 "don't touch the screen"

the place where to spawn the text using those fonts? i tryied with size=10,bottom=10 etc but nothing worked

do we really not have another solution?

3:58 AM

also it only happens on your device

Szybet

can i write to /tmp/ some data, sync and sleep 2 and then mem > state?

i guess?

If you're curious about the recoveryfs, here's the main scripts directory: https://github.com/Kobo-InkBox/recoveryfs/tree/master/opt/recovery/scripts

6:09 AM

The restore-*.sh ones are probably the most interesting

6:11 AM

btw, here's how it looks on the screen: https://youtu.be/KaBj7acHRHk?t=2129 (starts at 35:29)

Szybet

but its now reacting instant for the power button

good, but will it work on other devices?

(edited)

Also, I implemented this: https://github.com/Kobo-InkBox/gui-bundle/commit/d48fe45756321afdb2ea7040e606b54c7e5a328c

inkbox script: Implement auto-restart failsafe on non-zero exit cod...

tux-linux

also it only happens on your device

And its working only on my device, there is a IF n306

1

tux-linux

do we really not have another solution?

I prefer this that way

tux-linux

good, but will it work on other devices?

(edited)

Why not

Well, on other devices the FB might cause problems while it works on the Nia

the error message is only for the nia (edited)

1

Szybet

I prefer this that way

Ok, please, at least, write it correctly with a capital in the beginning of the sentence

1:59 PM

I'm curious; does Wi-Fi reconnection work when waking from sleep on your device with what you implemented?

tux-linux

I'm curious; does Wi-Fi reconnection work when waking from sleep on your device with what you implemented?

for now no, but i only need to add one if in the code

2:08 PM

i have anoother issue

Ok

Szybet

i have anoother issue

Yes?

the device blocks going to sleep no matter what, only reboot helps

You can't wake up?

i cant suspend it

It happens everytime?

only if i test it heavily

2:18 PM

i think

What's your code?

#!/bin/bash
# /tmp/sleep_error stores errors when going to sleep. For kobo nia, if missing permission error will show up it will try again
DEVICE=$(cat /opt/inkbox_device)

# Race condition; going to sleep
echo "false" > /tmp/sleep_standby # Idk if its needed

/opt/bin/fbink/fbink -k -f -q
/opt/bin/fbink/fbink -t regular=/etc/init.d/splash.d/fonts/resources/inter-b.ttf,size=20 "Sleeping" -m -M -q
echo "sleeping" > /tmp/sleep_status

go_sleep() {
    echo  "read file"
    cat /tmp/sleep_error
    rm -f /tmp/sleep_error 2>/dev/null
    sleep 8
    sync
    echo "going to sleep"
    echo "1" > /sys/power/state-extended
    sync
    sleep 2
    echo "mem" > /sys/power/state 2>/tmp/sleep_error 
    sleep 2
}

go_sleep

if [ "${DEVICE}" == "n306" ]; then
    # Separate if to avoid writing many times to the e-ink. idk about the syntax / look of the code here
    if grep -q "Operation not permitted" /tmp/sleep_error; then
        ### TO DO get the good position of it
        /opt/bin/fbink/fbink -m -q -B WHITE -t regular=/etc/init.d/splash.d/fonts/resources/inter-b.ttf,size=18,top=800,bottom=0,left=0,right=0 "don't touch the screen"
        /opt/bin/fbink/fbink -m -q -B WHITE -t regular=/etc/init.d/splash.d/fonts/resources/inter-b.ttf,size=14,top=870,bottom=0,left=0,right=0 "Nia can't go to sleep if it's touched"
    fi
    while grep -q "Operation not permitted" /tmp/sleep_error; do
        go_sleep
    done
fi

/etc/init.d/suspend_scripts/after_sleep.sh

2:28 PM

the part with while is not working

2:29 PM

if it gets there, it will not suspend, just loop through and there will be x fails in suspend_stats

2:29 PM

and things in go_sleep like echo and cat are my debugging things

Maybe you can just store the status of echo command in a var ${?}

will this help with the problem?

No

2:34 PM

But it will make more sense probably

okay, i will implement it after it will work

When you write to sysfs you don't need to sync Files there are not real files, they are linked directly to the kernel

sync is for the filesystem

Yeah, put it only once

okay

Hmm I would add something blocking that would prevent executing go_sleep if it's already running But maybe the while loop already prevents that

there is no &, so its only running once

You can use sh -x to help you debug

2:39 PM

It's really useful

what it does?

Try it

2:40 PM

sh -x script.sh

2:41 PM

Well, maybe more bash -x Btw, if possible, always use /bin/sh as it is faster than bash If you need arrays, etc. it's okay but ideally the faster it runs the better

tux-linux

Well, maybe more bash -x Btw, if possible, always use /bin/sh as it is faster than bash If you need arrays, etc. it's okay but ideally the faster it runs the better

thats helpfull

this is weird

2:47 PM

the script if it goes to this loop, it cant suspend itself (edited)

2:47 PM

but then i relaunch the whole script, it suspends normal (edited)

I don't understand

maybe now

2:52 PM

what does /tmp/sleep_mode?

Tells other scripts/services that it's going/in sleep mode IIRC It is necessary

and /tmp/sleep_standby?

Used for WiFi reconnection

WEIRD

3:28 PM

nothing makes sense

3:28 PM

i needed to add:

 kill -CONT $(pidof inkbox-bin 2>/dev/null) 2>/dev/null
        kill -CONT $(pidof oobe-inkbox-bin 2>/dev/null) 2>/dev/null
        kill -CONT $(pidof calculator-bin 2>/dev/null) 2>/dev/null
        kill -CONT $(pidof scribble 2>/dev/null) 2>/dev/null
        kill -CONT $(pidof lightmaps 2>/dev/null) 2>/dev/null
        kill -CONT $(pidof qreversi-bin 2>/dev/null) 2>/dev/null
        kill -CONT $(pidof 2048-bin 2>/dev/null) 2>/dev/null

        sleep 2

        kill -STOP $(pidof inkbox-bin 2>/dev/null) 2>/dev/null
        kill -STOP $(pidof oobe-inkbox-bin 2>/dev/null) 2>/dev/null
        kill -9 $(pidof lockscreen-bin 2>/dev/null) 2>/dev/null
        kill -STOP $(pidof calculator-bin 2>/dev/null) 2>/dev/null
        kill -STOP $(pidof scribble 2>/dev/null) 2>/dev/null
        kill -STOP $(pidof lightmaps 2>/dev/null) 2>/dev/null
        kill -STOP $(pidof qreversi-bin 2>/dev/null) 2>/dev/null
        kill -STOP $(pidof 2048-bin 2>/dev/null) 2>/dev/null

to make it go to sleep after the error with touch

3:28 PM

why?

Wait, where's the sleep call in this code?

where?

Szybet

i needed to add:

 kill -CONT $(pidof inkbox-bin 2>/dev/null) 2>/dev/null
        kill -CONT $(pidof oobe-inkbox-bin 2>/dev/null) 2>/dev/null
        kill -CONT $(pidof calculator-bin 2>/dev/null) 2>/dev/null
        kill -CONT $(pidof scribble 2>/dev/null) 2>/dev/null
        kill -CONT $(pidof lightmaps 2>/dev/null) 2>/dev/null
        kill -CONT $(pidof qreversi-bin 2>/dev/null) 2>/dev/null
        kill -CONT $(pidof 2048-bin 2>/dev/null) 2>/dev/null

        sleep 2

        kill -STOP $(pidof inkbox-bin 2>/dev/null) 2>/dev/null
        kill -STOP $(pidof oobe-inkbox-bin 2>/dev/null) 2>/dev/null
        kill -9 $(pidof lockscreen-bin 2>/dev/null) 2>/dev/null
        kill -STOP $(pidof calculator-bin 2>/dev/null) 2>/dev/null
        kill -STOP $(pidof scribble 2>/dev/null) 2>/dev/null
        kill -STOP $(pidof lightmaps 2>/dev/null) 2>/dev/null
        kill -STOP $(pidof qreversi-bin 2>/dev/null) 2>/dev/null
        kill -STOP $(pidof 2048-bin 2>/dev/null) 2>/dev/null

to make it go to sleep after the error with touch

i added this to while and now it goes to sleep, and does not loop itself

Well it makes no sense

but it works

3:44 PM

maybe some of those app tryied were awaken when the error accoured, and then tryied to get input from touch?

Without kill cont and only with kill stop, does it work?

nope

3:49 PM

do you have the full script?

nothing makes sens now

nothing in the kobo world ever makes sense, period.

i have. i will test some more things and then send it to you (edited)

can i see it now?

what part exactly?

the entire script

they are like 5 of them

well, those 5 then

3:50 PM

i want to understand why the kill thing works

4.01 KB

1.41 KB

1.79 KB

3.15 KB

1.22 KB

3:53 PM

i havent tested everything yet

3:53 PM

how do i clear half of the e-ink?

why?

?

Szybet

how do i clear half of the e-ink?

why?

becouse the wifi mark appears sometimes when launching the kill thing

look into fbink docs refresh rectangle or something

3:55 PM

fbink without arguments will print the full help

Szybet

Click to see attachment 🖼️

I am not sure I am ready to merge this It will need a LOT of testing, and it seems it affects other devices too It will be something to test for 2.0 release

it needs testing, sure

can't you put touch_issue_fix.sh into a function?

this will load it to ram, the code will be rarely used, the code in this file will be uglier

why uglier?

4:05 PM

and it will load the file to ram anyway

tux-linux

and it will load the file to ram anyway

really?

tux-linux

why uglier?

20 lines of kill

and? (edited)

well okay

Szybet

this will load it to ram, the code will be rarely used, the code in this file will be uglier

it will load <1K to RAM, so I guess it is negligible

oh, and idle sleep is broken

4:12 PM

i think

4:14 PM

why is the wifi logo even refreshing when the state didn't changed

i know it's set to refresh every 10 s

why ;_;

4:16 PM

can you change it?

4:16 PM

its better that if i did in my script the rectangle thing

i will try

good

4:18 PM

thanks

1

https://github.com/Kobo-InkBox/inkbox/commit/cfdace7dabef537989c56f51684343cb0da3e0bd also that was a failure, sorry

Reader: Disable setEnabled() calls on 'Next' and 'Previous' page bu...

Otherwise, the screen would flash everytime, regardless of the setting set by the user. I'm keeping the code there in case I or someone else find a fix.

4:21 PM

i tried everything but it didn't want to work

disabling buttons refreshes the screen?

yes

wtf qt

i know

i will add this rectangle anyway

4:24 PM

becouse battery / time has the same problem, and they change anyways

1

done

4:28 PM

https://github.com/Kobo-InkBox/inkbox/commit/e4a21f5c85e5d1079de6f8434bf801e1cb936746

Wi-Fi icon improvements · Kobo-InkBox/inkbox@e4a21f5

Don't update it if it isn't necessary.

nice

1

wifi recconection works

good

tux-linux

https://github.com/Kobo-InkBox/inkbox/commit/cfdace7dabef537989c56f51684343cb0da3e0bd also that was a failure, sorry

maybe dont disable the button, but give an if at the slot that its executing when clicked?

already tried

and?

But the function is blocking, so it always finishes executing before launching a new one, and it remembers at the same time how many times you clicked on the button. It's like a queue So it never worked

??? it can execute to the end, but you simply disable the code in it

4:58 PM

i dont understand something propably

it's very complicated

4:58 PM

i'm not sure I can even explain it

idle sleep works too

5:05 PM

everything works

5:09 PM

sleep_manager

426 bytes

suspend_scripts.zip

4.54 KB

5:09 PM

here is everything

ok; i don't really have the energy to test this right now but i'll consider it for 2.0 release thanks!

i will use it for me

1

I updated the GUI bundle Scrollbars should be better You'll also need to update the rootfs (edited)

Okay,i will test it today

6:44 AM

Im taking tommorow the nia for a trip and i hope it will be usable, it will not break himself

1

tux-linux

The GUI uses qtextedit

https://stackoverflow.com/questions/3254652/several-ways-of-placing-an-image-in-a-qtextedit

Several ways of placing an image in a QTextEdit

I think this is a very simple question, but when I copy an image I can't paste it in a QTextEdit? Paste is inactive! Also I would like to know how to drag-and-drop a picture. BTW I use the following

6:28 PM

umm?

You can??

6:32 PM

Ok, didn't know that The class name is misleading

idk, but here i see its possible

6:32 PM

i will use it in sanki for sure

1

Does the boot progress bar work correctly on your device?

um yes?

1

Also, are X11 apps launching/working fine?

Szybet

um yes?

Good, just wanted to check

tux-linux

Also, are X11 apps launching/working fine?

yes, but fbink gets errors after some time when x11 is enabled

Yeah sure

10:24 PM

That I can't do anything about

10:24 PM

A Qt terminal would be nice

yes!

Though we'd have to figure out how to get out of the GUI chroot

10:25 PM

Maybe telnet, but that is not a really good idea

10:26 PM

Because the GUI doesn't have a bundled SSH client

hmm... there must be a way to escape chroot

Yes there is But it does not make a very good impression if it's official hehe

10:30 PM

I think I have an idea

10:36 PM

Throw a dropbear static binary into /kobo/external_root/, SSH to localhost (main rootfs) with it, and bingo (edited)

this makes sense

10:37 PM

here is the whole new sleep management, i forgot to include one file

sleep.zip

5.81 KB

1

Seems that my Glo HD too wakes on touchscreen

well, the fix would now be easy

10:47 PM

i think

Depends if it's the same file

10:49 PM

The Glo HD has an IR touchscreen

IR?

The scroll bars are now usable, thanks

11:02 PM

except the brightness dialog

11:02 PM

the screen shifts even if they aren't any special characters in book names (edited)

Huh?

What is you question?

Szybet

IR?

Infrared?

Szybet

the screen shifts even if they aren't any special characters in book names (edited)

Question

11:28 PM

Yes

tux-linux

Yes

Interesting one. But the call to wake the device should be the same

tux-linux

Question

Well what can i say, thats what is happening to pe (edited)

11:49 PM

The file chooser and last books section need a complete redesign anyway, so dont care about it

Szybet

the screen shifts even if they aren't any special characters in book names (edited)

Screenshot?

Szybet

The file chooser and last books section need a complete redesign anyway, so dont care about it

It's going into 1.9 release at least, I care!

Szybet

Interesting one. But the call to wake the device should be the same

And it did not do anything :/

tux-linux

And it did not do anything :/

Are you sure? I needed to edit 3 Giles at least

well, i wiped out the whole 'wake' function, if that doesn't work then I don't know what works

what are other irq ( interrupt ) commands in those files?

Also, there is just one file for the zForce touchscreen

+ can you compile the driver as module

tux-linux

Also, there is just one file for the zForce touchscreen

Impossible

6:33 AM

+ can you unbind the driver on the device?

https://github.com/Kobo-InkBox/kernel/blob/master/kernel/linux-3.0.35-n437/drivers/input/touchscreen/zforce_i2c.c

Szybet

Impossible

why?

tux-linux

why?

Im optimistic

tux-linux

https://github.com/Kobo-InkBox/kernel/blob/master/kernel/linux-3.0.35-n437/drivers/input/touchscreen/zforce_i2c.c

I will look at it later

1

6:37 AM

+ you can gry the wake up interface i talked anout

11:31 AM

Thats something to fix, but its sleeping anyway so thats good

Did you try USB debug mode on the Nia?

You mean?

BOOT_USB_DEBUG

Yes, it works

1

nice

12:47 AM

kobo kernel code is a nightmare

12:48 AM

I still can't understand why it still resumes on touchscreen even if I tore the whole function down...

12:48 AM

diff --git a/kernel/linux-3.0.35-n437/drivers/input/touchscreen/zforce_i2c.c b/kernel/linux-3.0.35-n437/drivers/input/touchscreen/zforce_i2c.c
index a562702d9f..583126728c 100755
--- a/kernel/linux-3.0.35-n437/drivers/input/touchscreen/zforce_i2c.c
+++ b/kernel/linux-3.0.35-n437/drivers/input/touchscreen/zforce_i2c.c
@@ -712,28 +712,6 @@ static int zForce_ir_touch_suspend(struct platform_device *pdev, pm_message_t st
 
 static int zForce_ir_touch_resume(struct platform_device *pdev)
 {
-       if (gSleep_Mode_Suspend) {
-               if ( !(41==gptHWCFG->m_val.bPCB && 3!=gptHWCFG->m_val.bUIConfig) ) { //auto active while power on zforce controller
-                       if(8==gptHWCFG->m_val.bTouchCtrl || 11==gptHWCFG->m_val.bTouchCtrl) {
-                               i2c_master_send(zForce_ir_touch_data.client, cmd_Active_v2, sizeof(cmd_Active_v2));
-                       }else{
-                               i2c_master_send(zForce_ir_touch_data.client, cmd_Active, sizeof(cmd_Active));
-                       }       
-               }
-               if( custom_touchSizeLimit_set ) {
-                       printk("command setTouchSizeLimits");
-                       i2c_master_send(zForce_ir_touch_data.client, cmd_SetTouchSizeLimits, sizeof(cmd_SetTouchSizeLimits));
-                       msleep(5);
-               }
-//             enable_irq_wake(zForce_ir_touch_data.client->irq);
-               request_irq(zForce_ir_touch_data.client->irq, zForce_ir_touch_ts_interrupt, IRQF_TRIGGER_FALLING, ZFORCE_TS_NAME, ZFORCE_TS_NAME);
-       }
-       else
-               disable_irq_wake(zForce_ir_touch_data.client->irq);
-       if (!zForce_ir_touch_detect_int_level ()) {
-               zForce_ir_touch_ts_triggered ();
-       }
-       
        return 0;
 }

12:49 AM

I mean, if that doesn't work, I don't know what works

Me too, but look up to #chat

I will need to disable wake up for the charger too. When its 100% it wakes him up, then it goes to sleep but it wakes I'm up and it's a circle (edited)

1

I updated the build_u-boot.sh script in the kernel repo; building U-Boot is now much easier

1:33 AM

Example for the Nia:

env TOOLCHAINDIR=/home/build/inkbox/kernel/toolchain/arm-nickel-linux-gnueabihf TARGET=arm-nickel-linux-gnueabihf THREADS=16 scripts/build_u-boot.sh n306

It was easy already, but still cool

1

Can you fdisk -l your SD card?

Yes, i will do it today

tux-linux

Can you fdisk -l your SD card?

kobo:~# fdisk -l
Disk /dev/mmcblk0: 15 GB, 15931539456 bytes, 31116288 sectors
486192 cylinders, 4 heads, 16 sectors/track
Units: sectors of 1 * 512 = 512 bytes

Device       Boot StartCHS    EndCHS        StartLBA     EndLBA    Sectors  Size Id Type
/dev/mmcblk0p1    24,0,1      38,63,32         49152      79871      30720 15.0M 83 Linux
/dev/mmcblk0p2    51,0,1      550,63,32       104448    1128447    1024000  500M 83 Linux
/dev/mmcblk0p3    551,0,1     678,63,32      1128448    1390591     262144  128M 83 Linux
/dev/mmcblk0p4    679,0,1     857,31,32      1390592   31116287   29725696 14.1G 83 Linux
kobo:~#

why do you need it

9:01 PM

and: fuse-overlayfs 14G 2.7G 11G 20% /kobo so i have 11G available for books, its 3 more than on nickel with the same sd card

9:05 PM

the whole screen is shifting when there are recent books, with no special characters in recent books. removing files in kobo:/opt/config/08-recent_books# fixes the issue. Its needs a whole redesign, but i just say that its completly broken

Szybet

kobo:~# fdisk -l
Disk /dev/mmcblk0: 15 GB, 15931539456 bytes, 31116288 sectors
486192 cylinders, 4 heads, 16 sectors/track
Units: sectors of 1 * 512 = 512 bytes

Device       Boot StartCHS    EndCHS        StartLBA     EndLBA    Sectors  Size Id Type
/dev/mmcblk0p1    24,0,1      38,63,32         49152      79871      30720 15.0M 83 Linux
/dev/mmcblk0p2    51,0,1      550,63,32       104448    1128447    1024000  500M 83 Linux
/dev/mmcblk0p3    551,0,1     678,63,32      1128448    1390591     262144  128M 83 Linux
/dev/mmcblk0p4    679,0,1     857,31,32      1390592   31116287   29725696 14.1G 83 Linux
kobo:~#

why do you need it

dd if=/dev/mmcblk0 bs=512 count=49151 of=n306.bin

10:41 PM

Send me the file

what is this... ( EDIT: i deleted those files ) (edited)

10:46 PM

there is some interesting... text?

It's not supposed to weigh like 20M?

i coppied your command?

What's the output?

i have no idea why it was 8mb

11:17 PM

now its 25

Ok thanks

11:21 PM

It's for imgtool

?

11:22 PM

so?

11:22 PM

there is some text

11:22 PM

maybe a serial number

Unpartitioned space Contains waveform Yeah, I'll blank out the serial number

i hope you do it

Szybet

maybe a serial number

InkBox UID? Does it match the one in Settings?

?

Szybet

i hope you do it

InkBox UID is pretty much worthless regarding warranty or anything

Szybet

?

Open settings, show system info

tux-linux

InkBox UID is pretty much worthless regarding warranty or anything

It's a random string of 256 bytes

tux-linux

InkBox UID is pretty much worthless regarding warranty or anything

its something not from nickel?

tux-linux

Open settings, show system info

yes its the same

1

Yeah, it overwrites the serial number

good

11:27 PM

and the settings menu is broken

11:27 PM

there is a horizontal slider

Picture?

In what way is it broken?

11:29 PM

It's normal There's not enough space for some element

11:29 PM

(I suspect the timezone thing)

tux-linux

It's normal There's not enough space for some element

its bad

What do you want that I do

11:30 PM

It just doesn't fit the screen

timezone could cut out the text or move the box to the left, if its really it

tux-linux

It just doesn't fit the screen

its a horibble design and it can be fixed

I would need to remove the reboot req. text

11:31 PM

Which I'm sure you would say would be worse

tux-linux

I would need to remove the reboot req. text

no

Or reduce the qcombobox font size But I already did that and it's getting really small

11:34 PM

Maybe I can do it

where is the ui file for the seconf page

In the second page of settings.ui

11:34 PM

Don't reduce the font size of timezone label, it will break the design language

11:35 PM

The combo box already breaks it in a quiet way and I don't really like it

tux-linux

In the second page of settings.ui

what

Or maybe you have a better abbreviation for reboot required

a toast after setting timezone

Szybet

what

There are two arrows on the top right of the qstackedwidget

but its not a good fix

Szybet

a toast after setting timezone

Which would say what?

11:36 PM

Oooh

tux-linux

Which would say what?

what

But if then you mess up one time with the combo box and want to set it again, you have to wait

11:37 PM

That's why I put it in the label

11:37 PM

I can't figure out why it needs a reboot but it is needed

a toast that dissapears after a click?

Maybe

tux-linux

I can't figure out why it needs a reboot but it is needed

wait... on arch linux you need to source a file or smth to apply timezone

No, here it's a bind mount and symlink problem Nonsense, I didn't try to understand more

umount / mount? meh

That could be something

tux-linux

There are two arrows on the top right of the qstackedwidget

where...

Maybe adding that to ibxd and calling it when the combo box value changed could fix the problem

https://unix.stackexchange.com/questions/17051/how-to-change-timezone-to-utc-without-rebooting-aix maybe this: https://askubuntu.com/questions/1145565/should-i-reboot-after-set-new-timezone-using-timedatectl

Szybet

where...

Don't have access to that now Otherwise you can click the page_* elements in the right bar

Szybet

https://unix.stackexchange.com/questions/17051/how-to-change-timezone-to-utc-without-rebooting-aix maybe this: https://askubuntu.com/questions/1145565/should-i-reboot-after-set-new-timezone-using-timedatectl

It already works in the rootfs live

maybe you need to do this in chroot

It's just the bind mount in the Kobo chroot that doesn't point to the right thing

or restart inkbox_gui

You really think an user would like that?

after clicking ok applying settings and 10s wait for the gui

tux-linux

It's just the bind mount in the Kobo chroot that doesn't point to the right thing

It's like the bind mount records the content of the symlink and never updates it afterwards

its better than reboot

No, I think a simple unmount/mount in ibxd can work

tux-linux

Don't have access to that now Otherwise you can click the page_* elements in the right bar

thats some serious magic

11:44 PM

hm?

Also you need to get back to page 1 before saving the file or it will display the last page you were in in the compiled version

11:44 PM

No, the text just takes up too much space in the first place

11:44 PM

I think that fixing the reboot required problem will be better

tux-linux

No, the text just takes up too much space in the first place

cut it off

11:47 PM

use a combo widget that will use the entire screen

11:47 PM

really, anything

tux-linux

I think that fixing the reboot required problem will be better

I will do that

tux-linux

dd if=/dev/mmcblk0 bs=512 count=49151 of=n306.bin

is this actually a good idea? in never devices, some other revisions the chips / parts could be diffrent, so a diffrent config is there. it could damage or brick the device?

We'll find out when it happens

8:13 PM

InkBox is not providing any warranty

8:14 PM

So, not our fault if it breaks the device

tux-linux

We'll find out when it happens

it will be to late

tux-linux

InkBox is not providing any warranty

put a big warning, or make a check status

8:14 PM

dd this and that and compare it

Also, we don't have a choice If we don't include that unpartitioned space the device won't boot

Szybet

dd this and that and compare it

a guide, maybe i will write it

We need devices to compare

8:16 PM

But it is also a nightmare for Kobo to put a different unpartitioned space in each device So when they change hardware we will probably know

well okay

Besides, it's not like unpartitioned space from another device will brick it I have a touch B and put the Touch C image on it It didn't work well, but no problems and when using a correct U-Boot and kernel there was no permanent damage

well okay but its still risk

Waveform data is not portable

12:02 AM

the rest should be

NiLuJe

Waveform data is not portable

you mean?

(IIRC, I don't remember most of the crap in there ;p).

12:04 AM

waveform data is tied to a specific production series of screens

NiLuJe

waveform data is tied to a specific production series of screens

@tux-linux so we have a problem

There used to be a QR code or some shit on the flex cable to ID 'em

12:05 AM

its next to it

12:05 AM

in the nia

12:05 AM

propably

(This may or may not be less of an issue on NTX boards, but it was fatal (as in, broken refreshes) on Kindles

NiLuJe

waveform data is tied to a specific production series of screens

No, shit But how do you explain the people that got InkBox up and running without issues on their devices? How often does the production line change?

NiLuJe

(This may or may not be less of an issue on NTX boards, but it was fatal (as in, broken refreshes) on Kindles

Reversible?

12:54 AM

The wrong waveform on the Touch B did no harm at all after a clean SD flash

NiLuJe

There used to be a QR code or some shit on the flex cable to ID 'em

And how do you explain the success people have in replacing their screens without actually scanning the QR code and just going with the generic model number?

Which is why I said something something NTX ;p.

1:01 AM

It's an issue for sure on lab126 boards

Ok I'm just kind of freaking out right now...

And theoretically, every waveform is tied to a specific vcom

I discussed about that with @andi and came to the conclusion that it didn't matter, but apparently not (edited)

NiLuJe

And theoretically, every waveform is tied to a specific vcom

vcom?

I don't remember, brain is mush. it broke screen replacements on Kindles for sure

1

I had people on GH that tested the stock images with their device and they apparently didn't encounter problems either @dill also has it working on his Mini

Keep in mind NTX is probably outputting less boards than lab126 by a significant order of magnitude

1

1:05 AM

Making these kind of changes much harder to catch in the wild

1

tux-linux

I had people on GH that tested the stock images with their device and they apparently didn't encounter problems either @dill also has it working on his Mini

yep! working great

1

So @tux-linux maybe a script?

Why? How? It will make the installing experience crap Imagine running it on Windows

imagine using windows

2:07 PM

a script that will write the first half of the image and the other one, without overwriting waveform

How do you do that on Windows?

run wsl2 for it

There is no dd, I think you have to write in cpp or something an entire app

do you have a better - safe idea?

Szybet

run wsl2 for it

You really think everyone will be willing to install wsl2 on their PC just for that? Also you can access device nodes in wsl2? Didn't know that

i-dont-know

2:12 PM

thats just my idea

Maybe we should create a cross platform Qt app, InkBox image flasher, like the rpi one

2:12 PM

That would be my best guess

then how do you use it on windows? you said there is no dd

2:15 PM

also, script, app, whatever

You need to invent dd

2:16 PM

I guess

2:16 PM

Write raw bytes to the MMC

2:16 PM

Yeah, this will not be fun

nope nope nope nope xD

I think that's what etcher does

http://www.chrysocome.net/dd

2:18 PM

there are some dd for windows

Yes, but no sane user of Windows will be willing to open a terminal and enter commands and do something they haven't ever done before

2:20 PM

For now, I will put a warning, also the good thing is that the wrong waveform doesn't seem to brick the device, but the warning will be there

tux-linux

Yes, but no sane user of Windows will be willing to open a terminal and enter commands and do something they haven't ever done before

sadly

2:23 PM

they should make a copy of the image

2:23 PM

everyone

Yeah, but again it requires cli

2:26 PM

And imgtool is probably not that suited for Windows or any end user, it uses qemu-nbd and asks for root access every minute

2:53 PM

Fine?

describe the issue maybe more?

3:37 PM

waveform etc

3:37 PM

and backup

http://23.163.0.39/bundles/inkbox/native/n306/inkbox-1.9-n306.xz Please test if recoveryfs boots and gets to main GUI interface

Doesn't recovery was made by the update? (edited)

...no?

8:59 AM

Have you seen the actual size of the recoveryfs?

500 Mb

9:58 AM

can i just copy files from the image recovery partition and put then on the kobo?

And why recovery is bigger than rootfs

10:29 AM

Are there libraries for p4?

Szybet

can i just copy files from the image recovery partition and put then on the kobo?

yeah you could do that

Szybet

And why recovery is bigger than rootfs

because it contains user data + rootfs + kernel + u-boot

User data?

contents of p4

This makes more sense

tux-linux

http://23.163.0.39/bundles/inkbox/native/n306/inkbox-1.9-n306.xz Please test if recoveryfs boots and gets to main GUI interface

Does it boot to recoveryfs?

I will check it today

tux-linux

http://23.163.0.39/bundles/inkbox/native/n306/inkbox-1.9-n306.xz Please test if recoveryfs boots and gets to main GUI interface

2:45 PM

2:46 PM

7h~

2:48 PM

okay, now its better

2:48 PM

but still, it jumps like crazy between those speeds

It works

3:35 PM

thats a pretty cool setup

1

3:35 PM

maybe give a note for the nia + the other ereader that kobox can cause crashes

1

passcode code doesn't work

3:44 PM

i mean, while going to sleep

3:44 PM

when it boots its fine, im asked for it

Szybet

passcode code doesn't work

Yeah, that's sometimes to fix up for 2.0

Szybet

i mean, while going to sleep

this horrible sleep manager is back D:

tux-linux

Yeah, that's sometimes to fix up for 2.0

do i need to create a github issue or not for this?

No I won't forget to fix it

sleep doesn't work even when disabling passcode

3:47 PM

not cool

Szybet

sleep doesn't work even when disabling passcode

In what way?

oh, look: 1. disabled passcode 2. sleep button clicked 3. error 255 4. it goes back to inkbox 5. it goes to sleep, but in the inkbox menu so it seems like it freezes

Maybe try to reboot

3:49 PM

At least it goes to sleep

tux-linux

Maybe try to reboot

now it works (edited)

1

oh i forgot to test the think with plato

hmm

11:13 PM

in kobo:/kobo/sys/devices/platform/soc/2100000.aips-bus/21f8000.i2c/i2c-3/3-0032/ricoh619-battery/power_supply/mc13892_bat

11:13 PM

capacity            status              type
current_now         subsystem           uevent
device              technology          voltage_min_design
health              temp                voltage_now
power               time_to_empty_now
present             time_to_full_now

11:13 PM

time_to_empty_now seems to say something but full no

11:14 PM

technology says li-io, but in the device is a li po -_-

11:14 PM

stupid kobo

11:14 PM

temp says 35~ while charging?

11:15 PM

i should feel that heat

11:15 PM

health says good

11:16 PM

current doesn't change ;/

11:17 PM

are these informations available in other devices?

11:17 PM

uevent:

POWER_SUPPLY_NAME=mc13892_bat
POWER_SUPPLY_STATUS=Discharging
POWER_SUPPLY_PRESENT=1
POWER_SUPPLY_VOLTAGE_NOW=4136700
POWER_SUPPLY_CURRENT_NOW=895
POWER_SUPPLY_CAPACITY=93
POWER_SUPPLY_TEMP=340
POWER_SUPPLY_TIME_TO_EMPTY_NOW=30420
POWER_SUPPLY_TIME_TO_FULL_NOW=3932100
POWER_SUPPLY_TECHNOLOGY=Li-ion
POWER_SUPPLY_HEALTH=Good
POWER_SUPPLY_VOLTAGE_MIN_DESIGN=3350

11:18 PM

thats all interesting

11:18 PM

i will check in the future if its the same in nickel, becouse this is weird

The ttfull and ttempty do fluctuate almost sensibly, yeah

11:33 PM

(they're pretty fucking jittery, though, the simple act of sending a read that way will skew the results)

does li-ion matter? or should there be li-po

11:34 PM

^for battery life

Doesn't matter, but I don't recall it saying Li-Po anywhere

well in nickel battery life is better, so im searching the issue

11:35 PM

@tux-linux how do you launched plato? it has some lib problems

It's available as early as in Mk. 5, at least: https://svn.ak-team.com/svn/Configs/trunk/Kindle/Kobo_Hacks/Scripts/bat_ttl.sh

11:36 PM

Nickel aggressively puts the device into standby after ~5s of inactivity

11:36 PM

(actual suspend to ram w/ wakeup pins untouched)

11:37 PM

Shouldn't be overly noticeable on i.MX SoCs, though, but makes a huge difference on sunxi

NiLuJe

Nickel aggressively puts the device into standby after ~5s of inactivity

maybe thats why it was made that the touchscreen puts the device from sleep? (edited)

Que?

edited it, my fault

Wakes up, you mean?

yes

11:37 PM

interrupts

Yep

11:38 PM

buttons & touch

this makes sense

and hall sensor and gyro, for that matter

but how they did that when i put it properly to sleep, it doesn't wake up when i touch the screen? (edited)

We've just finished implementing that in KOReader

11:38 PM

state-extended to 1 disables the wakeup pins

but this didn't worked for the nia?...

shrug

11:39 PM

Don't have a Nia, but if it's the same kernel as the Libra 2, it's probably hilariously borked (edited)

its propably similar

i have tryied to get this error about e-ink for an hour, no success -_-

12:31 AM

im giving up for today

12:31 AM

i have no idea why, how

12:31 AM

i will try tommorow to get it

Did you enable x11?

Yes, of course

1:09 AM

I have masę various thing in various apps

1:09 AM

Nothing

Well, nice then Click the next button a thousand times when reading a book

1:09 AM

Toggle options menu several times a second Etc

I tryied

1:12 AM

I will tommorow too

tux-linux

Well, nice then Click the next button a thousand times when reading a book

No, not nice

1:13 AM

We still dont know the issue

Szybet

No, not nice

In the sense that you can probably use KoBox extensions

Propably

I was trying some more, nothing

12:14 AM

No error, kobox works

1

12:15 AM

Now we need to know why

There is no explanation When you want to show that sometime doesn't work, it suddenly works

12:16 AM

Suggestion: spam epdc with multiple fbink writes

I was having kobox errors on the self made inkbox 2 month ago, the differences between now and then were:

my self compiled kernel ( wakeups from the screen were still there
no recovery partition
USB ip, was on, but I turned it on tonight and it works
files were put be myself on the partition, maybe I missed something, maybe they were damaged because of network
what more... idk, maybe you know?

the only thing I would see would be the kernel but even there, it isn't related even in a remote way

Szybet

I was having kobox errors on the self made inkbox 2 month ago, the differences between now and then were:

my self compiled kernel ( wakeups from the screen were still there
no recovery partition
USB ip, was on, but I turned it on tonight and it works
files were put be myself on the partition, maybe I missed something, maybe they were damaged because of network
what more... idk, maybe you know?

Time on the device was not working properly

again, it's irqwake stuff ...maybe?

tux-linux

the only thing I would see would be the kernel but even there, it isn't related even in a remote way

Maybe I could try an old kernel

maybe it is the fix for the libra

12:21 AM

you could don't expect everything to work in GUI/system though

tux-linux

maybe it is the fix for the libra

Yeah, flash the image on it, there is probably your self made inkbox

12:23 AM

You could try plato when the error is still there

libra has an emmc so, not doable that easily

Idk, install in some way the system build from image script

Wi-Fi being enabled or the device being plugged to a power source are somewhat proven way to avoid the epdc issues on the Libra 2, FWIW

Nonsense How can that work

I would wager switching to the performance governor and/or raising the minimum cpu freq would, too

Ok, thanks for the insight

bus frequency and weird arm crap

1

NiLuJe

Wi-Fi being enabled or the device being plugged to a power source are somewhat proven way to avoid the epdc issues on the Libra 2, FWIW

LIbra 2 is imx6sll right?

yup: https://github.com/kobolabs/Kobo-Reader/commit/fb6846604f975420f6d621b132d104181b2b2f8f

libra2 · kobolabs/Kobo-Reader@fb68466

Nia is imx6ull maybe it can work too Libra original is imx6sll

Oh, the Nia runs an ULL, never mind me

You never know...

There's like, three of 'em in the wild, hard to tell

1

Still, thanks for the suggestion, it might work to "fix" this problem for the Libra original

2:46 AM

and maybe in the Nia with some luck

NiLuJe

Wi-Fi being enabled or the device being plugged to a power source are somewhat proven way to avoid the epdc issues on the Libra 2, FWIW

i was looking for the issue with wifi enabled and plugged into power source...

8:42 AM

well more testing is needed

NiLuJe

Wi-Fi being enabled or the device being plugged to a power source are somewhat proven way to avoid the epdc issues on the Libra 2, FWIW

i tested it, its true

11:36 PM

+ the wifi interface dissapears after that

11:36 PM

message.txt

2.11 KB

maybe just the modules were unloaded?

11:44 PM

idk

Szybet

i tested it, its true

it cant be reversed sadly

11:53 PM

setting performance governor avoids the issue

11:53 PM

but it was tested shortly

11:53 PM

and kobox apps are faster

11:54 PM

and i dont know if unsetting it after would change something

11:54 PM

hmm

11:54 PM

@tux-linux what launches kobox apps, what is called when they are closed? i want to add some code for testing, described a bit above

Hmmm Interesting /etc/init.d/xorg.sh

You should test it on the libra tok

update screen permission denied happened to me now while connected to usb + wifi

7:02 PM

i was developing sanki (edited)

7:02 PM

not good

7:02 PM

i will try one more time but with everything this + performance governor

Happened again, with performance governor

8:28 PM

I assume its random

Sad

we would only need to know what causes the error in the kernel

10:16 PM

:/

Needs verbose epdc logging enabled (edited)

how?

I don't do kernels, check what actually enables the gallen_log stuff (or whatever it's actually using) (edited)

2:28 AM

Might be a mix of actually sane upstream stuff and crazy ntx hacks

Szybet

the command to compile:

env GITDIR="${PWD}" TOOLCHAINDIR="${PWD}/toolchain/arm-nickel-linux-gnueabihf/" THREADS=$(($(nproc)*2)) TARGET=arm-nickel-linux-gnueabihf scripts/build_kernel.sh n306 root

the command to write: sudo dd if=zImage-root of=/dev/sdd bs=512 seek=81920

this here

12:18 PM

to compile and write

12:18 PM

i hope

sleep.zip

5.81 KB

so continuing the thread from #dev

9:30 PM

the kernel compiled from my arch machine, not debian container

9:30 PM

what the f*ck is wrong with all of this

Arch

9:37 PM

Couldn't find the error in the log from my phone

tux-linux

Arch

did you even understand what im talking?

Yes

9:39 PM

I didn't have success with Gentoo

9:39 PM

IIRC not Arch either

tux-linux

I didn't have success with Gentoo

you have problems too?

I build it on Debian

there is something messed up with permissions

9:41 PM

i created a user with sudo access, cloned the repo clean and it compiled on debian

9:41 PM

it didn't worked with root ( user ) (edited)

9:42 PM

also, can i test the kernel with emu?

9:42 PM

if it will work

No

9:52 PM

It needs the emu kernel

meh

9:59 PM

I updated the kernel

9:59 PM

and error code 5 appered

9:59 PM

wtf

10:00 PM

can i access anything? do i need to open the device? ;_;

10:02 PM

well help @tux-linux

Szybet

can i access anything? do i need to open the device? ;_;

Probably

ok, why did this happen

10:22 PM

?

I don't know

what

It would say that because your rootfs signature is invalid

10:22 PM

OH

10:23 PM

You forgot to make device nodes

10:23 PM

Am I right?

?...

10:23 PM

why isint it in the script

10:23 PM

WHY?

there is a script

why isin't it launched from the main script

For some reason I don't remember It could be included

10:23 PM

It is also written in the readme

oh it is in the readme

10:24 PM

so the dumb person here is me

10:24 PM

nice...

Well, you know the fix

Where is my ifixit kit

10:24 PM

Maybe a credit card can do

why isin't debug mode launched?

10:29 PM

its a rooted kernel

Because it can't launch (edited)

becouse

No devices nodes

this makes sense

10:35 PM

can't you simple check if files exist ( device nodes ) ?

I will add it to the main script

do i need to do it myself / create a github issie?

tux-linux

I will add it to the main script

for 100%?

10:44 PM

with what key do i need to sign the gui bundle?

10:45 PM

the nia one?

10:47 PM

its an update.isa

10:47 PM

weird

10:47 PM

what

so basically how do i update the gui bundle

11:05 PM

replace in update/update.isa

11:05 PM

now i remember

we reallty need this guide

11:23 PM

Szybet_bundle is propably outdated

11:23 PM

Szybet

Click to see attachment 🖼️

What's the problem?

its asking me to update

11:26 PM

propably becouse i set gui_bundle version to szybet_bundle

11:26 PM

¯\_(ツ)_/¯

11:26 PM

is there a way to disable this message?

What's in .inkbox?

11:27 PM

huh

Really strange

11:28 PM

What rootfs commit are you on?

11:29 PM

Try restart update_inkbox

11:29 PM

Then relaunch GUI

tux-linux

What rootfs commit are you on?

i cloned it a hour ago

11:30 PM

cp: cannot create regular file '/kobo/mnt/onboard/.adds/qt-linux-5.15.2-kobo/lib/libzip.so': Read-only file system
0 files (152.0 KiB) copied in 0.1 seconds (  0.0 KiB/s).

11:30 PM

well

Szybet

cp: cannot create regular file '/kobo/mnt/onboard/.adds/qt-linux-5.15.2-kobo/lib/libzip.so': Read-only file system
0 files (152.0 KiB) copied in 0.1 seconds (  0.0 KiB/s).

so now how do i add single libraries

Repackage bundle

i mean for user apps

11:33 PM

they cant do that

11:33 PM

kobo:(unknown)# service update_inkbox restart
 * Checking for available updates ...
Update skipped or no update available, aborting ...                                                           [ ok ]
ash: getcwd: No such file or directory
kobo:(unknown)#

Szybet

so now how do i add single libraries

Add them in a lib/ folder for each app, we can tweak ld library path to look there too (edited)

How

Well, for each app that needs a particular lib, we can put it in a folder where the app is /mnt/onboard/.apps/[app-isa-root]/lib

12:48 PM

I think that doing it with squashfs in .isa is simpler for signing

12:51 PM

The idea: 1. Put .isa app bundle in .apps 2. On next reboot a shell script detects if there are apps there. If yes it verifies the bundle and if successful creates a folder and mounts the app to it. If signature is invalid it might display an alert splash or a dialog (generalDialog) to tell the user.

12:52 PM

It will need to be run each time inkbox_gui starts, and maybe it can be an ibxd FIFO function to allow live updating

12:53 PM

Why did I write this here

stop doing the boot longer

12:54 PM

i don't accept this

When do you want to verify apps then

we cant verify them once?

This is the biggest security risk

and this overcomplicates things

You keep asking me for security risks, there's one now

like how will you manage storage for apps? i want to store decks with sanki

tux-linux

You keep asking me for security risks, there's one now

i see

For that you can create a folder in apps called [app-name_data]

12:58 PM

Please go to #dev

sometimes i put my nia to sleep for one day, i come back and its 50% sometimes i come back, its 99%

11:43 PM

is TP3(?)8 (the 4 pads next to the console) a uart?

5:17 PM

the bluetooth one? or the disabled one?

you mean?

5:18 PM

the big pads that are THT are uart, i use them

5:18 PM

but the small ones on the left, idk

yes, the non-THT

5:21 PM

https://github.com/Szybet/niAudio/blob/main/images/nia/img8.jpg -> what is the chip besides C137?

niAudio/img8.jpg at main · Szybet/niAudio

Kobo nia ereader with audio added. Contribute to Szybet/niAudio development by creating an account on GitHub.

andi

yes, the non-THT

well i didnt tryied tht ones are connected to the cpu, so i can see uboot messages

andi

https://github.com/Szybet/niAudio/blob/main/images/nia/img8.jpg -> what is the chip besides C137?

wait

https://github.com/Szybet/niAudio/blob/main/images/nia/img10.jpg the chip on the flex pcb?

niAudio/img10.jpg at main · Szybet/niAudio

Kobo nia ereader with audio added. Contribute to Szybet/niAudio development by creating an account on GitHub.

i wonder why i didn't uploaded it earlier to github

chips.pdf

7.55 MB

5:26 PM

here is everything you want to know

5:26 PM

clear pictures with names that i readed using a magnifying glass

one typo: Ricoh RC5... and not AC5

oh okay

BWNCSA : very interesting: I guess it is Sy7636A but on the Kobo Libra H2O, it has a different marking

well i uploaded everything on github

5:45 PM

this pdf

LAN eKTF2232ALW: LAN is part of the logo of elan, so it is probably eLAN

7:32 PM

we have

7:32 PM

andi@aktux:~/kernel/drivers/input/touchscreen$ grep 'compatible.*ektf' *.c ektf2127.c: { .compatible = "elan,ektf2127" }, ektf2127.c: { .compatible = "elan,ektf2132" }, elants_i2c.c: { .compatible = "elan,ektf3624", .data = (void *)EKTF3624 },

7:33 PM

hopefully it is compatible

it does look like it

That's the controller in the Sage & Elipsa, fwiw

7:33 PM

(It sucks hairy balls)

btw... isint it a risk running an untested kernel?

Very prone to throwing events into the void when it's in low power state

Szybet

btw... isint it a risk running an untested kernel?

hardware level damage?

the risk is not totally zero, but I think the risk of damaging something while fiddling with serial cable is also there. But I think the most important parts are tested (cpu frequency handling for example). Next risk is mixing up pins. There are some chances to mix something important up

7:40 PM

But I have blew up hardware because of bad kernels

too much voltage on regulators for example?

yes

thats bad

7:42 PM

i dont really can afford blowing my only ereader

do you have shell access?

7:42 PM

to the nia

i have access to everything

so you can dump regulator states

its possible, not like i knew how

for name in /sys/class/regulator/regulator.* ; do echo `cat $name/name` `cat $name/microvolts` `cat $name/state` ; done (edited)

7:47 PM

no

7:47 PM

messed up

for name in /sys/class/regulator/regulator.* ; do echo $(cat $name/name) $(cat $name/microvolts) $(cat $name/state) ; done

(edited)

use ```

7:48 PM

nope

7:48 PM

message.txt

3.04 KB

changed it

obo:~# for name in /sys/class/regulator/regulator.* ; do echo `cat $name/name` `cat $n
ame/microvolts`  `cat $name/state` ; done
cat: /sys/class/regulator/regulator.0/microvolts: No such file or directory
cat: /sys/class/regulator/regulator.0/state: No such file or directory
regulator-dummy
vdd3p0 3200000 enabled
RICOH619_DC5 1800000 enabled
RICOH619_LDO1 900000 disabled
RICOH619_LDO2 900000 enabled
RICOH619_LDO3 900000 disabled
RICOH619_LDO4 900000 disabled
RICOH619_LDO5 600000 enabled
RICOH619_LDO6 600000 enabled
RICOH619_LDO7 900000 enabled
RICOH619_LDO8 1800000 enabled
RICOH619_LDO9 900000 disabled
cpu 1250000 enabled
RICOH619_LDO10 900000 disabled
RICOH619_LDORTC1 1700000 enabled
RICOH619_LDORTC2 900000 disabled
cat: /sys/class/regulator/regulator.23/microvolts: No such file or directory
DISPLAY_SY7636 disabled
cat: /sys/class/regulator/regulator.24/microvolts: No such file or directory
VCOM_SY7636 disabled
cat: /sys/class/regulator/regulator.25/microvolts: No such file or directory
V3P3_SY7636 disabled
cat: /sys/class/regulator/regulator.26/state: No such file or directory
TMST_SY7636 0
vddsoc 1250000 enabled
cat: /sys/class/regulator/regulator.4/state: No such file or directory
vref-3v3 3300000
cat: /sys/class/regulator/regulator.5/state: No such file or directory
SD1_SPWR 3000000
RICOH619_DC1 1400000 enabled
RICOH619_DC2 600000 enabled
RICOH619_DC3 600000 enabled
RICOH619_DC4 1200000 enabled
kobo:~#

7:50 PM

well i can help you prepare the kernel, but i wont test it

7:50 PM

tux-linux would be sad if he lost a contributor

7:50 PM

then we can wait for others to test it

Szybet

tux-linux would be sad if he lost a contributor

happens all the time hehe

7:51 PM

but you two seem to be hanging tight

here were some people that had the nia... but we cant ping them because we still didnt created roles for devices

tux-linux

happens all the time hehe

but its not a reason to do it intentionally knowing the risk

7:51 PM

It's been an incredibly unpopular device in my experience, FWIW

7:52 PM

So good luck finding actual users? ;o)

here were 5 people using it

7:52 PM

for sure

yeah

there were also like 3 people in xcsoar project

7:53 PM

maybe they would want to do this? they have problems with the existing kernel

(Which... makes sense, actually, it's a failed Clara 2)

7:53 PM

Hence the actual Clara 2 that's in the works ;p

7:53 PM

Possibly not an NTX board (here's hoping) (edited)

NiLuJe

Possibly not an NTX board (here's hoping) (edited)

it could be worse...

7:55 PM

thinks about B300

what are the options?

7:55 PM

i hope they wont lock it down

Well, from where I'm standing the latest round of i.MX devices are more of a failure than the sunxi ones (edited)

7:56 PM

We knew going in that the sunxi would be awful at power efficiency

7:56 PM

Because we're used to that on PB devices (edited)

7:56 PM

But at least they don't randomly crash every 5 seconds (lookin' at you, Libra 2)

NiLuJe

We knew going in that the sunxi would be awful at power efficiency

they need to put a bigger as for capacity battery in them... they are really small

That's the Elipsa. It goes by purely by having two gigantic cells.

7:57 PM

The Sage... not so much ;o)

well, I like the imx6 ones

And the hastily designed workaround (a.k.a., the PowerCover) behaves infinitely worse than a dumb usb power brick, it's hilarious

quite well pm in mainline kernel

Oh, definitely, I'd have been happy w/ an i.MX7D

that one has afaik bad pm in mainline kernel

7:59 PM

hmm, looking at ebay, the nia is at >100 euros...

7:59 PM

ouch

I'll believe it, AFAIK, rM had trouble with it, and lab126 too, it took them ages to polish it for the Oasis 2 (edited)

8:00 PM

Possibly longer than it took them to deal with a new platform for the PW5 (mtk), although they had extensive experience with that one with the Fires

8:00 PM

And it trounces the Sage, it's hilarious

8:01 PM

Probably helps that the kernel & hardware team doesn't feel like a lone guy working in his garage (hi, ntx ;p)

yeah

I think if someone makes good pm for imx7, then everyone will have fun with it

8:02 PM

it is imho more a task of NXP to mainline it

Yup

8:03 PM

Putting one out on aarch64 would be neat, too.

About the nia price, I expected something well below 100 euros

NiLuJe

Putting one out on aarch64 would be neat, too.

yes

8:03 PM

with 8G of RAM

8:03 PM

ok no I'm just dreaming

Although admittedly I have no idea how the power efficiency compares to our old cortex cores, so I don't really know if that would even make sense

1

andi

About the nia price, I expected something well below 100 euros

it was 500 zł ( 110 Euros ) when i buyed it almost a year ago

here in Canada we sometimes have very good deals about Kobos

in poland only one online shop selled one kobo, it was the nia...

do they sell tolinos?

no, Kobos only (AFAIK, eBay is not impossible though) (edited)

8:07 PM

Kobo's home country is Canada

8:07 PM

so, there you go

well, they question was more for @Szybet (edited)

oh

8:08 PM

it's probably possible to order them from Germany

andi

do they sell tolinos?

nope, i havent seen any but in germany they are everywhere, even in random bool libraries to buy

yes, a friend of mine bought one when there was a good deal. We thought it would be compatible with my Kobo Clara HD, everyone says they are identical

8:10 PM

looked at the chips, and board it seems to be

8:10 PM

but

8:11 PM

really looking at the SoC reveals that the shine3 has a imx6sl and not a 6sll

8:11 PM

the kobo bootloader managed to spit something out on serial/

andi

really looking at the SoC reveals that the shine3 has a imx6sl and not a 6sll

that's not identical at all...

but did not find the sdcard

probably PCB issue

8:12 PM

wrong memory addresses

but if pictures are bad you are too lazy to compare

8:12 PM

yes

8:12 PM

PCB seems same

8:12 PM

same markings

8:13 PM

too lazy compare exactly

Szybet

obo:~# for name in /sys/class/regulator/regulator.* ; do echo `cat $name/name` `cat $n
ame/microvolts`  `cat $name/state` ; done
cat: /sys/class/regulator/regulator.0/microvolts: No such file or directory
cat: /sys/class/regulator/regulator.0/state: No such file or directory
regulator-dummy
vdd3p0 3200000 enabled
RICOH619_DC5 1800000 enabled
RICOH619_LDO1 900000 disabled
RICOH619_LDO2 900000 enabled
RICOH619_LDO3 900000 disabled
RICOH619_LDO4 900000 disabled
RICOH619_LDO5 600000 enabled
RICOH619_LDO6 600000 enabled
RICOH619_LDO7 900000 enabled
RICOH619_LDO8 1800000 enabled
RICOH619_LDO9 900000 disabled
cpu 1250000 enabled
RICOH619_LDO10 900000 disabled
RICOH619_LDORTC1 1700000 enabled
RICOH619_LDORTC2 900000 disabled
cat: /sys/class/regulator/regulator.23/microvolts: No such file or directory
DISPLAY_SY7636 disabled
cat: /sys/class/regulator/regulator.24/microvolts: No such file or directory
VCOM_SY7636 disabled
cat: /sys/class/regulator/regulator.25/microvolts: No such file or directory
V3P3_SY7636 disabled
cat: /sys/class/regulator/regulator.26/state: No such file or directory
TMST_SY7636 0
vddsoc 1250000 enabled
cat: /sys/class/regulator/regulator.4/state: No such file or directory
vref-3v3 3300000
cat: /sys/class/regulator/regulator.5/state: No such file or directory
SD1_SPWR 3000000
RICOH619_DC1 1400000 enabled
RICOH619_DC2 600000 enabled
RICOH619_DC3 600000 enabled
RICOH619_DC4 1200000 enabled
kobo:~#

yes, these voltages in the nia look quite different

about the battery connectors in the nia and elsewhere: do you know where to order them or part numbers? The reason I am asking: I would like to do things like current measurement (especially during suspend, e.g. without the need so acces /sys/class/power_supply/*battery/current_now) with a stable setup without testing the protection circuit of the battery every day (edited)

just solder cables or ask on reddit

1:57 PM

on the nia i dont think it works

kobo:/sys/class/power_supply# cat /sys/class/power_supply/*/current_now
845
kobo:/sys/class/power_supply#

its always 845, even if i turn wifi on

Every device running my kernel has shown sane values so far

well it also shows, just doesnt change

because the driver there is a mess

¯\_(ツ)_/¯

1:59 PM

as everything

1:59 PM

maybe disconnect the battery / supply power over usb?

1:59 PM

as what niluje talked, it should work

yes, it works

2:03 PM

but for example that messy code turns off the fuel gauge in some situations, I would like to understand the effects, especially on power consumption, and there I have no other chance.

2:04 PM

if those connectors are not found, probably the easiest is to hack the libra

2:04 PM

as the battery has no connectors there anyways

2:07 PM

well, after adding some more kapton tape around the solderig points...

well maybe just use the connector from the battery

2:49 PM

cut the wires

2:49 PM

if you really want

I've been doing power draw measurements with a cheap USB meter (because my devices are closed), and it's accurate enough to at least confirm that sunxi is terrible at PM ;o)

11:33 PM

Because, yeah, the sysfs stuff reports wonky shit on NTX kernels, can confirm

11:35 PM

It's... less helpful on devices that refuse to enter suspend while connected to a power source, though ;o) (Mk < 6, IIRC?)

yes, and I want to be able to close my devices also, so I would rather use that plug solution. But for my Libra I would insert something permanently because I do not need to close it because of broken display. Power from battery or from usb is a different situation, so you will have limitied results. Interesting would be the analysis of this: https://github.com/jwrdegoede/rtl8189ES_linux/issues/82 but hopefully the better values of near-mainline kernels will already help here

card regulator is not turned off if iface is down · Issue #82 · jwr...

The sdio card is not properly unpowered when device is down. Expectation would be that regulator is powered down when iface isis down. Removing the module helps. pmos-shine3:/sys/bus/platform/devic...

Re: that issue; there's an sdio_wifi_power kernel module on most NTX boards that does just that

11:37 PM

It's... fairly hilarious actually, because it's just

11:38 PM

#include <linux/module.h>
#include <linux/init.h>
#include <linux/slab.h>

extern void ntx_wifi_power_ctrl (int isWifiEnable);
static int __init sdio_wifi_pwr_init(void)
{
        ntx_wifi_power_ctrl(1);
        return 0;
}
module_init(sdio_wifi_pwr_init);

static void __exit sdio_wifi_pwr_exit(void)
{
        ntx_wifi_power_ctrl(0);
}
module_exit(sdio_wifi_pwr_exit);

MODULE_DESCRIPTION("sdio wifi power control driver");
MODULE_AUTHOR("Netronix");
MODULE_LICENSE("GPL");

11:38 PM

That happens to be an ntx_io ioctl (edited)

11:38 PM

Someone finally realized that it was dumb as hell, and they stopped shipping that and started actually using the ioctl

is this a way to shut off the wifi chip completly?

https://github.com/koreader/koreader/blob/615072c7771ca801bbc1d70154956fac1d1b4b61/platform/kobo/disable-wifi.sh#L62-L66

koreader/disable-wifi.sh at 615072c7771ca801bbc1d70154956fac1d1b4b6...

An ebook reader application supporting PDF, DjVu, EPUB, FB2 and many more formats, running on Cervantes, Kindle, Kobo, PocketBook and Android devices - koreader/disable-wifi.sh at 615072c7771ca801b...

sooo yes?

#define CM_WIFI_CTRL 208

11:40 PM

(ntx_io)

i speak english, not c unfortunately

Also #define CM_BLUETOOTH_PWR 126 because same chip

i assume it as a yes

NiLuJe

Also #define CM_BLUETOOTH_PWR 126 because same chip

same chip has bluetooth?...

IIRC, Cervantes had some comment in their HAL about it on some of their boards, where they purposefully chose to NOT power it off because they didn't like the latency trade-off

11:43 PM

That was paired with the usual "we sleep here because ntx told us to" comment around module loading/unloading that anyone that's ever dealt with wifi on ntx should be familiar with

(edited)

Szybet

i speak english, not c unfortunately

ioctl on /dev/ntx_io

tux-linux

ioctl on /dev/ntx_io

yea and

to explain what you seemed not to understand

Szybet

is this a way to shut off the wifi chip completly?

???

well, probably, according to KOReader

are we implementing it?

no

12:04 AM

it works now

12:04 AM

so

12:04 AM

fine

12:04 AM

i just need to fix the Glo HD

12:04 AM

but that's probably unrelated

tux-linux

it works now

"battery problems on the nia"

12:05 AM

did you heard that?

oh, didn't see

12:05 AM

well try it

tux-linux

well try it

https://github.com/Kobo-InkBox/inkbox/discussions/38

Future To-Do's and possible feature requests · Discussion #38 · Kob...

Turning off the interface using ioctl on the nia and other devices: in toggle.sh and ipd

12:22 AM

there is already too much to do...

NiLuJe

Re: that issue; there's an sdio_wifi_power kernel module on most NTX boards that does just that

well, that all is silly, userspace should have one way for all devices. That is rfkill block wifi

8:30 AM

and a good kernel should hide most platform-specific stuff

8:31 AM

if you accept removing modules as a solution, rmmod 8189fs

kobo good kernel those things dont combine

does the thing on mainline

so removing modules actually shuts the chip off? no need for this ioctl?

that is why I am trying to mainline things as much as possible

8:31 AM

on mainline kernels, yes

8:32 AM

the ioctl is simply not there

8:32 AM

on ntx kernels, you can rmmod sdio_wifi_power

oh okay

8:33 AM

goos that we already do that

8:33 AM

so why koreader has this?

perhaps some older ntx kernels need that

no idea, as said I prefer using newer stuff

@tux-linux maybe not thr nia, but older devices need that?

root@akepd:/sys/bus/sdio/drivers/rtl8189fs# ls -l total 0 --w------- 1 root root 4096 Aug 31 08:35 bind lrwxrwxrwx 1 root root 0 Aug 31 08:35 mmc1:0001:1 -> ../../../../devices/platform/soc/2100000.bus/2198000.mmc/mmc_host/mmc1/mmc1:0001/mmc1:0001:1 --w------- 1 root root 4096 Aug 31 08:34 uevent --w------- 1 root root 4096 Aug 31 08:35 unbind root@akepd:/sys/bus/sdio/drivers/rtl8189fs# echo mmc1:0001:1 >unbind root@akepd:/sys/bus/sdio/drivers/rtl8189fs# echo mmc1:0001:1 >bind root@akepd:/sys/bus/sdio/drivers/rtl8189fs# root@akepd:/sys/bus/sdio/drivers/rtl8189fs#

8:36 AM

that also works on mainline

it doesnt crash? cool

8:36 AM

is it maybe possible to turn off voltage regulators for maximum power efficency?

the regulator for the wifi is turned off

8:37 AM

in that case

the rest

8:37 AM

they are many

that is to be researched, and regulators have to be tied to the devices using them

too complicated why arm isint using chipsets that handle that?

also complicated

8:39 AM

giving a schematic

8:39 AM

and then we know what the regulators are doing

8:40 AM

what they are for

and I can properly add that information to the devicetree

&usdhc3 {
        pinctrl-names = "default", "state_100mhz", "state_200mhz", "sleep";
        pinctrl-0 = <&pinctrl_usdhc3>;
        pinctrl-1 = <&pinctrl_usdhc3_100mhz>;
        pinctrl-2 = <&pinctrl_usdhc3_200mhz>;
        pinctrl-3 = <&pinctrl_usdhc3_sleep>;
        vmmc-supply = <&reg_wifi>;
        mmc-pwrseq = <&wifi_pwrseq>;
        cap-power-off-card;
        non-removable;
        status = "okay";

Here i am specifying the power supply for the wifi card. Then the host controller driver will turn off regulator when the card is not used. If no driver for the card is loaded, it is considered not used. A well behaved driver (!= rtl8189fs out-of-mainline-tree driver) will also tell when the card is not used, so the host controller will know what to do

9:03 AM

so here are the information which are in that sdio_wifi_power module or some other dirty ntx stuff on many ntx kernels

what's up with tp36?

12:09 AM

secondary serial port?

you mean?

that group of 4 pads next to the console

never tested really

TP3-TP6: They are connected with the cable to the touchscreen, so probably some i2c around there

12:15 AM

can be detected by running i2cdetect on the corresponding bus

12:16 AM

and just looking at them with a multimeter

12:16 AM

i2cdetect in a loop

On the clara there is another i2c bus which is more difficult to access on pullups (without zeros on it) near the lm3630a, the situation might be the same on the nia. Hot candidates are R139, R143, TP26/25, less hot R146, TP84

1:01 AM

but turn off the backlight during any experiments here

1:01 AM

or measurements

1:01 AM

since the backlight has a comparatively high voltage

1:03 AM

lm3630a = U11

TP182-TP285: UART with HW-handshake hof non-soldered BT chip?

yep

TP36 really begs for attaching something to it

2:16 PM

using some spring contacts, so without soldering so things keep modular

2:18 PM

lots of distance to any other victim

2:19 PM

so just some backcover with holes in that place is needed

2:20 PM

My guess would be that UART4 is there

how to find out: 1. check whether there are signs that pins are used by something else

2:32 PM

can be checked via devicetree

2:32 PM

then check that they are configured as gpio

2:33 PM

and then play around with enabling pullup/-down ressistors

2:33 PM

via pad configuration in the SoC

2:33 PM

and check if a) gpio input moves, b) voltage on the pins is affected

2:34 PM

that in coarse

could be done

so uart4 would be gpio64/65 (LCD_ENABLE/LCD_CLK)

3:02 PM

in devicetree if used as gpio: &gpio3 0 &gpio3 1

3:02 PM

no signs of that

uart2 is &gpio4 6/7, aka gpio102/103 aka NAND_DATA0[45]

do you have the devicetree?

11:11 PM

of it

11:11 PM

extracted?

?

11:11 PM

its in the kernel source?

no, hidden space

11:11 PM

before 1. partition

11:12 PM

extract-kobohidden clara-bkup.img 1286 out.dtb

you mean this? https://github.com/Szybet/niAudio/blob/main/old/logs/ntx_hwconfig-more

niAudio/ntx_hwconfig-more at main · Szybet/niAudio

Kobo nia ereader with audio added. Contribute to Szybet/niAudio development by creating an account on GitHub.

https://github.com/akemnade/kobo-firmware-extractor

GitHub - akemnade/kobo-firmware-extractor: extracts firmware from h...

extracts firmware from hidden partitions on Kobo/Tolino ebook readers - GitHub - akemnade/kobo-firmware-extractor: extracts firmware from hidden partitions on Kobo/Tolino ebook readers

oh, cool

11:18 PM

well if you dont want to wait, there is the nia image for inkbox

11:18 PM

i can tommorow / in the weekend extract it

http://23.163.0.39/bundles/inkbox/native/n306/inkbox-1.9-n306.xz

if it contains your dtb, then I can extract it

Yes it's Szybet's unpartitioned space

11:19 PM

I don't have a Nia

reason: Kuratius' Nia has U67 soldered and that makes me extremely excited

???

the qfn thing which is not soldered on your board

What does it do

WHAAAT

11:20 PM

D:

11:21 PM

which chip is it

and its not soldered to his too?

it is soldered

andi

Click to see attachment 🖼️

i dont see a chip in this red square???

it is an excerpt from your image

Szybet

Click to see attachment 🖼️

but thats kuratios image

or just an optic illusion

what???...

so, no not soldered there too

11:31 PM

the pcb looks darker

11:32 PM

and so I got the impression that the chip might be there

11:32 PM

would be too nice

What does it do

in the device tree is something about bluetooth

has anyone ever compiled a kernel on this device? (edited)

on the nia, not for the nia?

yes

6:14 PM

on the nia

why

6:14 PM

no

6:14 PM

sounds as a horrible idea

I sometimes to such things for stability tests

6:17 PM

/*

on i.MX6ULL, no seperated VDD_ARM_IN and VDD_SOC_IN,

to align with other platform and use the same cpufreq

driver, still use the seperated OPP define for arm

and soc.

/ operating-points = < / kHz uV */ 900000 1250000 792000 1200000 528000 1175000 396000 1175000 198000 1175000 >;

6:17 PM

I found that comment and that makes no sense

yes, there are no separate inputs but there are two separate internal regulators, so you have separate voltages

6:30 PM

and the datasheet says for 900Mhz, typ. 1.275 V and not 1.250V, so again dear NTX what are you doing here?!

andi

I sometimes to such things for stability tests

shouldnt a compiler / crosscompiler always give the same results, without looking what cpu its running on?

yes, it should but it is often a good stability test

6:36 PM

there are several reports where some "internal gcc errors" were caused by hw stability issues

6:36 PM

I have also seen random error messages

stability as in hardware / software load test?

yes

oh, okay

6:36 PM

well those are ereaders, not regular pc

6:37 PM

its not designed with this in mind

6:37 PM

maybe even, depending on time something could burn off

6:37 PM

so i have concerns

well, maybe I am a bit more aggressive here, but you could also have some unintentional load test by opening a too big book, or some web page hogging the cpu when using the internal browser, so I would expect that a device would sustain that.

my point is time

6:56 PM

no one tested if an ereader can sustain 30h of full load on everything

6:57 PM

maybe temperature of one of the components is slowly going up

well, I do not urge you to try that thing, and I would not do in the hot sun

i had this problem on an asus tablet, luckly it was the cpu so it just shutted down

Szybet

maybe even, depending on time something could burn off

No

7:33 PM

My Kobo hardware has sustained heavy loads more times than it would ever have wanted to and it's still working

for how many h

Days?

7:40 PM

It can throttle itself if needed

7:40 PM

Didn't catch on fire

tux-linux

It can throttle itself if needed

cpu, yes, other components, not really

I remember poeple building a lot of stuff on-device back in the Kindle Touch days, too, FWIW (edited)

andi

yes, these voltages in the nia look quite different

well the regulator driver takes the voltages from a cache, which is never initialized if regulator is probed via devicetree (which is the case with the nia), so we are seeing the lowest end of the possible voltage range. The cache is in that case only set if the voltage is explicitely set by the kernel and not kept at boot state.

8:31 AM

Nasty surprises with ntx kernels everywhere... I hate that (edited)

extract-kobohidden nia2-bkup.img 1286 nia2-factory.dtb

tux-linux changed the channel name: Kobo Nia model A (N306A) 7/13/2023 9:47 PM